SB 1011: Energy: Utility Infrastructure AI Safety, Oversight, and Workforce Protection Act.
- Session Year: 2025-2026
- House: Senate
- Latest Version Date: 2026-05-18
Current Status:
In Progress
(2026-05-28: Ordered to inactive file on request of Senator McNerney.)
Introduced
In Committee
Awaiting First Chamber Vote
Passed First Chamber
In Committee
Awaiting Second Chamber Vote
Passed Second Chamber
Enacted
Existing law vests the Public Utilities Commission with regulatory jurisdiction over public utilities, including electrical corporations and gas corporations, while local publicly owned electric utilities are under the direction of their governing boards. Existing law requires every public utility to furnish and maintain adequate, efficient, just, and reasonable service, instrumentalities, equipment, and facilities, as are necessary to promote the safety, health, comfort, and convenience of its customers, its employees, and the public.
This bill would require the commission, on or before January 1, 2028, to adopt standards for an electrical or gas corporations use of artificial intelligence models, as provided. The bill would require the commission to direct an electrical or gas corporation to file a plan that demonstrates the corporations compliance with those standards. The bill would authorize the commission to prohibit an electrical or gas corporations use of an artificial intelligence model if the commission finds that deployment of the artificial intelligence model would negatively impact the provision of safe, affordable, and reliable electrical or gas service.
The bill would require each community choice aggregator or and local publicly owned electric utility to adopt a policy regarding its use of an artificial intelligence model that is consistent with the standards.
Under existing law, a violation of an order, decision, rule, direction, demand, or requirement of the commission is a crime.
Because a violation of a commission action implementing those requirements would be a crime, this bill would impose a state-mandated local program. Additionally, by imposing new duties on local publicly owned electric utilities, the bill would impose a state-mandated local program.
The California Constitution requires the state to reimburse local agencies and school districts for certain costs mandated by the state. Statutory provisions establish procedures for making that reimbursement.
This bill would provide that no reimbursement is required by this act for specified reasons.
Existing law vests the Public Utilities Commission (PUC) with regulatory jurisdiction over public utilities, including electrical corporations and gas corporations (privately owned utilities), corporations, while local publicly owned electric utilities and local publicly owned gas utilities (publicly owned utilities) are under the direction of their governing boards. Existing law requires the State Energy Resources Conservation and Development Commission (Energy Commission) to oversee the implementation of certain programs, including the California Renewables Portfolio Standard Program, by local publicly owned electric utilities. Under existing law, a violation of an order, decision, rule, direction, demand, or requirement of the PUC is a crime. every public utility to furnish and maintain adequate, efficient, just, and reasonable service, instrumentalities, equipment, and facilities, as are necessary to promote the safety, health, comfort, and convenience of its customers, its employees, and the public.
This bill would require the PUC, for a privately owned utility, and the Energy Commission, for a publicly owned utility, to oversee the implementation of a specified program to regulate automated decision systems in connection with certain utility functions. The bill would require privately owned utilities and publicly owned utilities (covered utilities) that employ automated decision systems in the mapping, design, configuration, operation, maintenance, or oversight of electrical or gas infrastructure to maintain a structured process by which qualified personnel are able to modify or override the output of the automated decision systems and to take other specified actions. The bill would prohibit a covered utility from deploying a high-risk automated decision system in its live operational environment unless it files with the PUC or Energy Commission, as appropriate, a safety plan containing certain information, and would require the high-risk automated decision system to operate in staging mode, as provided, before full operational deployment. The bill would require a high-risk automated decision system that creates, modifies, updates, or purports to correct system records to meet certain requirements. The bill would require a covered utility to report to the PUC or Energy Commission, as appropriate, within 24 hours of discovering any event in which a high-risk automated decision system contributed to or caused certain consequences, including a service interruption or outage affecting more than 500 customers, and would require the covered utility, within 30 days of the event, to submit a root-cause report to the PUC or Energy Commission, as appropriate, that includes certain information. The bill would require a covered utility to continuously monitor its high-risk automated decision systems and to submit an annual report to the PUC or Energy Commission, as appropriate, with certain information. The bill would require a covered utility to provide at least 180 days advance notice, as provided, to affected labor organizations and employees in impacted employee classifications before introducing any technological change involving automated decision systems that materially affects job duties, classifications, staffing levels, or training, and to develop retraining programs, as specified. The bill would prohibit a covered utility from implementing a high-risk automated decision system in its operations that results in the layoff of certain employees unless the covered utility has first exhausted any feasible retraining, redeployment, or reclassification options. The bill would subject a privately owned utility violating its requirements to enforcement pursuant to specified laws. Because the bill would subject a privately owned utility to those specified laws, and because a violation of a PUC action implementing the bills requirement would be a crime, the bill would impose a state-mandated local program. The bill would require a publicly owned utility to annually certify to its governing board and the Energy Commission its compliance with the bills requirements and regulations, guidelines, or procedures adopted to implement the bills requirements. By imposing additional duties on local publicly owned electric utilities and local publicly owned gas utilities, the bill would impose a state-mandated local program.This bill would require the commission, on or before January 1, 2028, to adopt standards for an electrical or gas corporations use of artificial intelligence models, as provided. The bill would require the commission to direct an electrical or gas corporation to file a plan that demonstrates the corporations compliance with those standards. The bill would authorize the commission to prohibit an electrical or gas corporations use of an artificial intelligence model if the commission finds that deployment of the artificial intelligence model would negatively impact the provision of safe, affordable, and reliable electrical or gas service.
The bill would require each community choice aggregator or local publicly owned electric utility to adopt a policy regarding its use of an artificial intelligence model that is consistent with the standards.
Under existing law, a violation of an order, decision, rule, direction, demand, or requirement of the commission is a crime.
Because a violation of a commission action implementing those requirements would be a crime, this bill would impose a state-mandated local program. Additionally, by imposing new duties on local publicly owned electric utilities, the bill would impose a state-mandated local program.
The California Constitution requires the state to reimburse local agencies and school districts for certain costs mandated by the state. Statutory provisions establish procedures for making that reimbursement.
This bill would provide that no reimbursement is required by this act for specified reasons.
Existing law vests the Public Utilities Commission (PUC) with regulatory jurisdiction over public utilities, including electrical corporations and gas corporations (privately owned utilities), while local publicly owned electric utilities and local publicly owned gas utilities (publicly owned utilities) are under the direction of their governing boards. Existing law requires the State Energy Resources Conservation and Development Commission (Energy Commission) to oversee the implementation of certain programs, including the California Renewables Portfolio Standard Program, by local publicly owned electric utilities. Under existing law, a violation of an order, decision, rule, direction, demand, or requirement of the PUC is a crime.
This bill would require the PUC, for a privately owned utility, and the Energy Commission, for a publicly owned utility, to oversee the implementation of a specified program to regulate automated decision systems in connection with certain utility functions. The bill would require privately owned utilities and publicly owned utilities (covered utilities) that employ automated decision systems in the mapping, design, configuration, operation, maintenance, or oversight of electrical or gas infrastructure to maintain a structured process by which qualified personnel are able to modify or override the output of the automated decision systems and to take other specified actions. The bill would prohibit a covered utility from deploying a high-risk automated decision system in its live operational environment unless it files with the PUC or Energy Commission, as appropriate, a safety plan containing certain information, and would require the high-risk automated decision system to operate in staging mode, as provided, before full operational deployment. The bill would require a high-risk automated decision system that creates, modifies, updates, or purports to correct system records to meet certain requirements. The bill would require a covered utility to report to the PUC or Energy Commission, as appropriate, within 24 hours of discovering any event in which a high-risk automated decision system contributed to or caused certain consequences, including a service interruption or outage affecting more than 500 customers, and would require the covered utility, within 30 days of the event, to submit a root-cause report to the PUC or Energy Commission, as appropriate, that includes certain information. The bill would require a covered utility to continuously monitor its high-risk automated decision systems and to submit an annual report to the PUC or Energy Commission, as appropriate, with certain information. The bill would require a covered utility to provide at least 180 days advance notice, as provided, to affected labor organizations and employees in impacted employee classifications before introducing any technological change involving automated decision systems that materially affects job duties, classifications, staffing levels, or training, and to develop retraining programs, as specified. The bill would prohibit a covered utility from implementing a high-risk automated decision system in its operations that results in the layoff of certain employees unless the covered utility has first exhausted any feasible retraining, redeployment, or reclassification options. The bill would subject a privately owned utility violating its requirements to enforcement pursuant to specified laws. Because the bill would subject a privately owned utility to those specified laws, and because a violation of a PUC action implementing the bills requirement would be a crime, the bill would impose a state-mandated local program. The bill would require a publicly owned utility to annually certify to its governing board and the Energy Commission its compliance with the bills requirements and regulations, guidelines, or procedures adopted to implement the bills requirements. By imposing additional duties on local publicly owned electric utilities and local publicly owned gas utilities, the bill would impose a state-mandated local program.
The California Constitution requires the state to reimburse local agencies and school districts for certain costs mandated by the state. Statutory provisions establish procedures for making that reimbursement.
This bill would provide that no reimbursement is required by this act for specified reasons.
This act shall be known, and may be cited, as the Utility Infrastructure AI Safety, Oversight, and Workforce Protection Act.
SEC. 2.(a)The Legislature finds and declares all of the following:
(1)The electrical and gas infrastructure of the state is vital to public safety, economic stability, and environmental sustainability.
(2)The increasing use of artificial intelligence models within utility operations, including mapping, design, configuration, control, maintenance, and oversight, introduces new risks to reliability, safety, asset integrity, system-of-record accuracy, and workforce continuity.
(3)Licensed engineers, technical specialists, and field operations personnel possess professional judgment essential to safe utility operations. Artificial intelligence models should complement, not replace, that judgment.
(4)Workers in utility operations should have appropriate notice and training opportunities when technological change affects their roles.
(b)In enacting this act, it is the intent of the Legislature to establish consistent statewide standards for the safe, transparent, auditable, and equitable implementation of artificial intelligence models in utility infrastructure operations.
SEC. 3.Chapter 11 (commencing with Section 8510) is added to Division 4.1 of the Public Utilities Code, to read:SEC. 3.Chapter 12 (commencing with Section 8510) is added to Division 4.1 of the Public Utilities Code, to read:11.12.Artificial Intelligence Models in Utility Infrastructure 8510.(a)For purposes of this chapter,
(b)Artificial intelligence model means an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments.
(c)Electrical corporation has the same meaning as defined in Section 218.
(d)Gas corporation has the same meaning as defined in Section 222.
(e)Local publicly owned electric utility has the same meaning as defined in Section 224.3.
8511.(a)As part of a new or existing proceeding, the commission shall, on or before January 1, 2028, adopt standards for an electrical corporations or gas corporations use of artificial intelligence models. The standards adopted by the commission shall do at least all of the following:
(1)Establish disclosure requirements for an electrical corporation or gas corporation to specify the types of artificial intelligence models it uses and how those models are used by the electrical corporation or gas corporation.
(2)Require an electrical corporation or gas corporation to identify the employee job classifications impacted by any planned implementation of an artificial intelligence model and efforts taken to ensure effective education, training, and retention for impacted employees.
(3)Identify steps an electrical corporation or gas corporation shall take to ensure that artificial intelligence models do not impact utility safety, affordability, and reliability.
(4)Establish requirements for human review and approval for the deployment and use of artificial intelligence models to ensure that electrical and gas services are safe, affordable, and reliable, including requirements for the placement of human review in the deployment of artificial intelligence models.
(5)Ensure that deployment of artificial intelligence models do does not displace utility employees needed for the safe, affordable, and reliable provision of electrical and gas services. For any job classification identified as responsible for human review and approval of an artificial intelligence model, the electrical corporation or gas corporation shall demonstrate that its workforce plan maintains sufficient personnel in that job classification to ensure that human review and approval processes are not rendered nominal by understaffing. the electrical corporation or gas corporation shall maintain staffing levels sufficient to support meaningful human review and approval processes, as determined by the commission.
(6)Establish requirements for human review and approval of any artificial intelligence model that does any of the following:(A)Makes or directly implements operational decisions affecting the mapping, design, configuration, operation, maintenance, or oversight of electrical or gas infrastructure without a mandatory human approval step integrated into the operational workflow before implementation.(B)Generates recommendations that the electrical corporation or gas corporation has configured to be automatically implemented without case-by-case human review.(C)Performs any act that could directly and foreseeably result in physical harm to a person or a service interruption or outage, or impact on public safety.(b)The requirements for human review and approval of artificial intelligence models in subdivision (a) shall ensure staff conducting the review and approval meet all of the following criteria:(1)Have relevant expertise in the operational area affected by the artificial intelligence model.(2)Have been provided sufficient information, time, and access to data to meaningfully evaluate the artificial intelligence models output or recommendations.(3)Retain unimpeded authority to reject, modify, or defer implementation of the artificial intelligence model without adverse consequence to the reviewer.(4)Have not been subject to a workflow design, performance metric, or operational pressure that renders rejection of the system output functionally impractical.(6)Identify and establish requirements for human review and approval of any use of artificial intelligence that is high risk, safety sensitive, or would have a material operational impact.
(c)(b)The commission may prohibit an electrical corporations or gas corporations use of an artificial intelligence model if the commission finds that deployment of the artificial intelligence model would negatively impact the provision of safe, affordable, and reliable electrical or gas service.
(d)(c)The commission shall direct an electrical corporation or gas corporation to file a plan that demonstrates the corporations compliance with the standards adopted pursuant to this section. The commission may request records as it deems necessary to verify compliance with the standards adopted pursuant to this section.
(e)(d)The commission shall ensure that a plan filed by an electrical corporation pursuant to subdivision (d) (c) may be reviewed by the Office of Energy Infrastructure Safety to ensure that any artificial intelligence model used for wildfire mitigation is consistent with the electrical corporations wildfire mitigation plan developed pursuant to Section 8386.
(f)(e)The commission shall ensure that impacted bargaining units are consulted in the development of plans filed pursuant to subdivision (d). (c).
(g)(f)(1)Each community choice aggregator established pursuant to Section 366.2 shall adopt a policy regarding its use of an artificial intelligence model.
(2)A policy adopted pursuant to this subdivision shall be consistent with the standards adopted by the commission pursuant to subdivision (a).
(h)(g)(1)Each local publicly owned electric utility shall adopt a policy regarding the its use of an artificial intelligence model.
(2)A policy adopted pursuant to this subdivision shall be consistent with the standards adopted by the commission pursuant subdivision (a).
8515.(a)This chapter establishes minimum safety and governance standards of artificial intelligence models.
(b)This chapter does not limit, waive, or alter any rights, remedies, or obligations under state or federal law, including the National Labor Relations Act (29 U.S.C. Sec. 151 et seq.), the Meyers-Milias-Brown Act (Chapter 10 (commencing with Section 3500) of Division 4 of Title 1 of the Government Code), the Ralph C. Dills Act (Chapter 10.3 (commencing with Section 3512) of Division 4 of Title 1 of the Government Code), or any collective bargaining agreement, with respect to technological change, staffing, workload, training, or working conditions.
8519.The provisions of this chapter are severable. If any provision of this chapter or its application is held invalid, that invalidity shall not affect other provisions or applications that can be given effect without the invalid provision or application.
SEC. 4.No reimbursement is required by this act pursuant to Section 6 of Article XIIIB of the California Constitution because a local agency or school district has the authority to levy service charges, fees, or assessments sufficient to pay for the program or level of service mandated by this act or because costs that may be incurred by a local agency or school district will be incurred because this act creates a new crime or infraction, eliminates a crime or infraction, or changes the penalty for a crime or infraction, within the meaning of Section 17556 of the Government Code, or changes the definition of a crime within the meaning of Section 6 of Article XIIIB of the California Constitution.
This act shall be known, and may be cited, as the Utility Infrastructure AI Safety, Oversight, and Workforce Protection Act.
SEC. 2.(a)The Legislature finds and declares all of the following:
(1)The electrical and gas infrastructure of the state is vital to public safety, economic stability, and environmental sustainability.
(2)The increasing use of automated decision systems artificial intelligence models within utility operations, including mapping, design, configuration, control, maintenance, and oversight, introduces new risks to reliability, safety, asset integrity, system-of-record accuracy, and workforce continuity.
(3)Licensed engineers, technical specialists, and field operations personnel possess professional judgment essential to safe utility operations. Automated decision systems Artificial intelligence models should complement, not replace, that judgment.
(4)Workers in utility operations should have appropriate notice and training opportunities when technological change affects their roles.
(b)In enacting this act, it is the intent of the Legislature to establish consistent statewide standards for the safe, transparent, auditable, and equitable implementation of automated decision systems artificial intelligence models in utility infrastructure operations.
SEC. 3.Chapter 11 (commencing with Section 8510) is added to Division 4.1 of the Public Utilities Code, to read:11.Automated Decision Systems Artificial Intelligence Models in Utility Infrastructure 8510.For purposes of this chapter, all of the following definitions apply:(a)(1)Automated decision system means a computational process, including one derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, and that is used to assist or replace human discretionary decisionmaking and materially impacts natural persons.(2)Automated decision system does not include any of the following:(A)Spam email filters.(B)Firewalls.(C)Antivirus software.(D)Identity and access management tools.(E)Calculators.(F)Databases, datasets, or other compilations of data.(b)Covered privately owned utility means either of the following:(1)An electrical corporation, as defined in Section 218.(2)A gas corporation, as defined in Section 222.(c)Covered publicly owned utility means either of the following:(1)A local publicly owned electric utility, as defined in Section 224.3.(2)A publicly owned gas utility.(d)Covered utility means either of the following:(1)A covered privately owned utility.(2)A covered publicly owned utility.(e)Engineering decision means a decision, recommendation, or configuration change affecting the design, operation, maintenance, or configuration of electrical circuits, substations, gas pipelines, compressor stations, or other utility infrastructure that is subject to review by, or is required to be performed by, a person licensed under the Professional Engineers Act (Chapter 7 (commencing with Section 6700) of Division 3 of the Business and Professions Code).(f)High-risk automated decision system means an automated decision system used by a utility that does any of the following: (1)Creates or modifies system records, including geographic information system (GIS) layers, asset registers, and configuration logs, used in operational or engineering decisionmaking.(2)Issues, prepopulates, or prioritizes operational actions, such as switching, fault isolation, circuit reconfiguration, load transfers, or gas-flow adjustments.(3)Prioritizes safety critical or wildfire risk mitigation decisions, including, any of the following:(A)Line deenergization or vegetation clearance.(B)The classification or escalation of suspected gas leaks or electrical hazards.(C)Recommendations affecting whether or when emergency response or field personnel dispatch is initiated.(4)Generates, prioritizes, or recommends either of the following:(A)Supervisory control and data acquisition.(B)Control room or operational control actions that alter operating states, including set points, valve lineups, compressor controls, pressure or flow adjustments, alarms, interlocks, or operating limits.(5)Creates, modifies, or recommends changes to safety critical system-of-record data that governs utility operations, including asset configuration data, isolation or valve lists, lockout or tagout points or procedures, alarm rationalization data, management of change records, or integrity-related operating parameters. (g)Provenance log means a traceable record that identifies the artificial intelligence model used, its version, its training data source, the time the model output was generated, the identity of any human reviewer, and the actions taken that are approved, modified, or rejected by the human reviewer.(h)Staging mode means the operating mode of a system where changes proposed by an artificial intelligence system are held in a nonoperational environment pending human review, testing, or validation before deployment into the live system. (i)System records means design documents, GIS layers, configuration logs or files, change-management records, mapping files, asset registers, or other digital or analog records used by a covered utility to design, plan, configure, operate, maintain, or oversee electrical or gas systems.(j)Technological change means the introduction of new or altered technology, equipment, software, automation, or robotics, or a new or altered artificial intelligence application, that alters the type, manner, or amount of work performed by employees of a covered utility.(k)Utility mapping system means a GIS or other digital platform used by a covered utility to record, monitor, design, configure, or manage the layout, configuration, status, or changes of utility infrastructure circuits, pipelines, substations, valves, or other equipment or assets.(l)Vendor automated decision system supply chain disclosure means the documentation provided by a vendor of an automated decision system that identifies model family, version, training data sources, update cadence, data residency, and known limitations or biases.8511.The commission shall oversee the implementation of this chapter by a covered privately owned utility. The Energy Commission shall oversee the implementation of this chapter by a covered publicly owned utility. The commission and Energy Commission shall coordinate their actions pursuant to this chapter to ensure that the requirements for covered privately owned utilities and covered publicly owned utilities are consistent.8512.(a)Every covered utility that employs an automated decision system in the mapping, design, configuration, operation, maintenance, or oversight of electrical or gas infrastructure shall maintain a structured process by which qualified personnel are able modify or override the output of the automated decision systems.(b)A covered utility shall not implement the output of a high-risk automated decision system without prior affirmative human review and approval of that output.(c)If the output of a high-risk automated decision system constitutes an engineering decision, the human review and approval required by subdivision (b) shall be performed by a California-licensed professional engineer. The high-risk automated decision system shall function as a decision-support tool and shall not independently execute operational actions or modify system records.(d)The covered utility shall maintain a provenance log for all automated decision systems outputs and human review actions, including timestamps, model version, training data source, human approvers identity, modifications and overrides, and outcome of the review. 8513.(a)A covered utility shall not deploy a high-risk automated decision system in its live operational environment unless it first files with the commission or Energy Commission, as appropriate, a safety plan that includes all of the following:(1)Information about the high-risk automated decision system that includes all of the following:(A)The model version.(B)Training data description.(C)Update cadence.(D)Known limitations.(E)Bias testing.(F)Cybersecurity controls.(G)Human-override provisions.(H)Roll-back features.(I)Logging capability.(2)Scenario testing results, including, at minimum, one wildfire or hazard scenario and one distributed energy resource-congestion or equipment-failure scenario, showing system behavior, failsafe behavior, human override, and system recovery.(3)A plan for monitoring the performance, periodic validation, and retraining or decommissioning of high-risk automated decision systems if performance falls below safety thresholds.(b)Before full deployment in a covered utilitys operation, the high-risk automated decision system shall operate in staging mode for a minimum of 18 months, or a commission-determined or Energy Commission-determined time period, as appropriate, with all changes audited and human approved before deployment to a live operational environment.(c)A covered utility shall provide a copy of the safety plan required by subdivision (a) to affected labor organizations upon filing or within 15 days upon request.8514.(a)A high-risk automated decision system that creates, modifies, updates, or purports to correct system records, including GIS layers, asset registers, and configuration logs, of a covered utility shall do all of the following:(1)Operate initially in staging mode until human review is complete.(2)Store proposed changes in a separate environment clearly labeled proposed by ADS pending human review and only be pushed to operational use after human approval.(3)Maintain a provenance log capturing model, version, training data source, human reviewer identity, review timestamp, modification approval or rejection, and system version history.(4)Include a roll-back mechanism enabling restoration of prior versions of system records in the event of a downstream adverse effect or error detection.(b)The covered utility shall maintain audit trails of all changes stemming from automated decision systems and human review for a minimum of ____ years and make the audit trails available to the commission or Energy commission, as appropriate, upon request.8515.(a)A covered utility shall report to the commission or Energy Commission, as appropriate, within 24 hours of discovering any event in which a high-risk automated decision system contributed to, or caused, any of the following:(1)A service interruption or outage affecting more than 500 customers.(2)Equipment damage, a failure, or a safety hazard.(3)A misswitching event, incorrect mapping or asset register update, erroneous configuration change, or data-integrity breach with safety or reliability implications.(b)Within 30 days of the occurrence of an event specified in subdivision (a), a root-cause report shall be submitted to the commission or Energy Commission, as appropriate, that includes both of the following:(1)The role of the high-risk automated decision system, vendor model and version of the high-risk automated decision system used, training data source, human oversight steps, and provenance log.(2)Corrective actions taken, scheduling of remediation, human-override review of the high-risk automated decision system, and actionable changes to safeguards.(c)(1)A covered privately owned utility shall continuously monitor its high-risk automated decision systems and submit an annual report to the commission on its risk-assessment update, performance versus baseline, any human-override statistics, and any near-misses or incidents.(2)A covered publicly owned utility shall continuously monitor its high-risk automated decision systems and submit an annual report to the Energy Commission and to its governing board on its risk-assessment update, performance versus baseline, any human-override statistics, and any near-misses or incidents.8516.(a)A covered utility shall provide at least 180 days advance notice to affected labor organizations and impacted employee classifications before introducing any technological change involving automated decision systems that materially affects job duties, classifications, staffing levels, or required training.(b)The notice shall include all of the following:(1)A description of the proposed technology, its intended function, and a timeline for deployment.(2)The anticipated or actual impacts on work processes, staffing, skill sets, and employee classifications.(3)A plan for retraining, redeployment, or reassignment of affected employees.(4)A demonstration that the proposed technology will not erode the work of licensed engineers, designers, or technical employees, unless retraining and deployment are offered.(c)Covered utilities shall develop joint retraining programs, in consultation with employee representatives, to ensure employees can transition to new work roles created by the technological change.(d)A covered utility shall not implement a high-risk automated decision system deployment in its operation that results in layoff of employees engaged in engineering decisionmaking, mapping, design, or technical operations unless the covered utility has first exhausted feasible retraining, redeployment, or reclassification options. 8517.The commission and Energy Commission may adopt regulations, guidelines, or procedures, as appropriate, to implement this chapter.8518.(a)A covered privately owned utility that violates this chapter may be subject to enforcement by the commission under its authority over safe and reliable operations and is subject to penalties under Section 2108, Section 2110, or other applicable provisions of this code.(b)A covered publicly owned utility shall annually certify to its governing board and Energy Commission that it is in compliance with this chapter and any adopted regulations, guidelines, or procedures implementing this chapter.8510.(a)For purposes of this chapter,
(b)Artificial intelligence model means an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments.
(c)Electrical corporation has the same meaning as defined in Section 218.
(d)Gas corporation has the same meaning as defined in Section 222.
(e)Local publicly owned electric utility has the same meaning as defined in Section 224.3.
8511.(a)As part of a new or existing proceeding, the commission shall, on or before January 1, 2028, adopt standards for an electrical corporations or gas corporations use of artificial intelligence models. The standards adopted by the commission shall do at least all of the following:
(1)Establish disclosure requirements for an electrical corporation or gas corporation to specify the types of artificial intelligence models it uses and how those models are used by the electrical corporation or gas corporation.
(2)Require an electrical corporation or gas corporation to identify the employee job classifications impacted by any planned implementation of an artificial intelligence model and efforts taken to ensure effective education, training, and retention for impacted employees.
(3)Identify steps an electrical corporation or gas corporation shall take to ensure that artificial intelligence models do not impact utility safety, affordability, and reliability.
(4)Establish requirements for human review and approval for the deployment and use of artificial intelligence models to ensure that electrical and gas services are safe, affordable, and reliable, including requirements for the placement of human review in the deployment of artificial intelligence models.
(5)Ensure that deployment of artificial intelligence models do not displace utility employees needed for the safe, affordable, and reliable provision of electrical and gas services. For any job classification identified as responsible for human review and approval of an artificial intelligence model, the electrical corporation or gas corporation shall demonstrate that its workforce plan maintains sufficient personnel in that job classification to ensure that human review and approval processes are not rendered nominal by understaffing.
(6)Establish requirements for human review and approval of any artificial intelligence model that does any of the following:
(A)Makes or directly implements operational decisions affecting the mapping, design, configuration, operation, maintenance, or oversight of electrical or gas infrastructure without a mandatory human approval step integrated into the operational workflow before implementation.
(B)Generates recommendations that the electrical corporation or gas corporation has configured to be automatically implemented without case-by-case human review.
(C)Performs any act that could directly and foreseeably result in physical harm to a person or a service interruption or outage, or impact on public safety.
(b)The requirements for human review and approval of artificial intelligence models in subdivision (a) shall ensure staff conducting the review and approval meet all of the following criteria:
(1)Have relevant expertise in the operational area affected by the artificial intelligence model.
(2)Have been provided sufficient information, time, and access to data to meaningfully evaluate the artificial intelligence models output or recommendations.
(3)Retain unimpeded authority to reject, modify, or defer implementation of the artificial intelligence model without adverse consequence to the reviewer.
(4)Have not been subject to a workflow design, performance metric, or operational pressure that renders rejection of the system output functionally impractical.
(c)The commission may prohibit an electrical corporations or gas corporations use of an artificial intelligence model if the commission finds that deployment of the artificial intelligence model would negatively impact the provision of safe, affordable, and reliable electrical or gas service.
(d)The commission shall direct an electrical corporation or gas corporation to file a plan that demonstrates the corporations compliance with the standards adopted pursuant to this section. The commission may request records as it deems necessary to verify compliance with the standards adopted pursuant to this section.
(e)The commission shall ensure that a plan filed by an electrical corporation pursuant to subdivision (d) may be reviewed by the Office of Energy Infrastructure Safety to ensure that any artificial intelligence model used for wildfire mitigation is consistent with the electrical corporations wildfire mitigation plan developed pursuant to Section 8386.
(f)The commission shall ensure that impacted bargaining units are consulted in the development of plans filed pursuant to subdivision (d).
(g)(1)Each community choice aggregator established pursuant to Section 366.2 shall adopt a policy regarding its use of an artificial intelligence model.
(2)A policy adopted pursuant to this subdivision shall be consistent with the standards adopted by the commission pursuant to subdivision (a).
(h)(1)Each local publicly owned electric utility shall adopt a policy regarding the its use of an artificial intelligence model.
(2)A policy adopted pursuant to this subdivision shall be consistent with the standards adopted by the commission pursuant subdivision (a).
8519.8515.(a)This chapter establishes minimum safety and governance standards of automated decision systems. artificial intelligence models.
(b)This chapter does not limit, waive, or alter any rights, remedies, or obligations under state or federal law, including the National Labor Relations Act (29 U.S.C. Sec. 151 et seq.), the Meyers-Milias-Brown Act (Chapter 10 (commencing with Section 3500) of Division 4 of Title 1 of the Government Code), the Ralph C. Dills Act (Chapter 10.3 (commencing with Section 3512) of Division 4 of Title 1 of the Government Code), or any collective bargaining agreement, with respect to technological change, staffing, workload, training, or working conditions.
8520.8519.The provisions of this chapter are severable. If any provision of this chapter or its application is held invalid, that invalidity shall not affect other provisions or applications that can be given effect without the invalid provision or application.
SEC. 4.No reimbursement is required by this act pursuant to Section 6 of Article XIIIB of the California Constitution because a local agency or school district has the authority to levy service charges, fees, or assessments sufficient to pay for the program or level of service mandated by this act or because costs that may be incurred by a local agency or school district will be incurred because this act creates a new crime or infraction, eliminates a crime or infraction, or changes the penalty for a crime or infraction, within the meaning of Section 17556 of the Government Code, or changes the definition of a crime within the meaning of Section 6 of Article XIIIB of the California Constitution.
This act shall be known, and may be cited, as the Utility Infrastructure AI Safety, Oversight, and Workforce Protection Act.
SEC. 2.(a)The Legislature finds and declares all of the following:
(1)The electrical and gas infrastructure of the state is vital to public safety, economic stability, and environmental sustainability.
(2)The increasing use of automated decision systems within utility operations, including mapping, design, configuration, control, maintenance, and oversight, introduces new risks to reliability, safety, asset integrity, system-of-record accuracy, and workforce continuity.
(3)Licensed engineers, technical specialists, and field operations personnel possess professional judgment essential to safe utility operations. Automated decision systems should complement, not replace, that judgment.
(4)Workers in utility operations should have appropriate notice and training opportunities when technological change affects their roles.
(b)In enacting this act, it is the intent of the Legislature to establish consistent statewide standards for the safe, transparent, auditable, and equitable implementation of automated decision systems in utility infrastructure operations.
SEC. 3.Chapter 11 (commencing with Section 8510) is added to Division 4.1 of the Public Utilities Code, to read:11.Automated Decision Systems in Utility Infrastructure 8510.For purposes of this chapter, all of the following definitions apply:
(a)(1)Automated decision system means a computational process, including one derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, and that is used to assist or replace human discretionary decisionmaking and materially impacts natural persons.
(2)Automated decision system does not include any of the following:
(A)Spam email filters.
(B)Firewalls.
(C)Antivirus software.
(D)Identity and access management tools.
(E)Calculators.
(F)Databases, datasets, or other compilations of data.
(b)Covered privately owned utility means either of the following:
(1)An electrical corporation, as defined in Section 218.
(2)A gas corporation, as defined in Section 222.
(c)Covered publicly owned utility means either of the following:
(1)A local publicly owned electric utility, as defined in Section 224.3.
(2)A publicly owned gas utility.
(d)Covered utility means either of the following:
(1)A covered privately owned utility.
(2)A covered publicly owned utility.
(e)Engineering decision means a decision, recommendation, or configuration change affecting the design, operation, maintenance, or configuration of electrical circuits, substations, gas pipelines, compressor stations, or other utility infrastructure that is subject to review by, or is required to be performed by, a person licensed under the Professional Engineers Act (Chapter 7 (commencing with Section 6700) of Division 3 of the Business and Professions Code).
(f)High-risk automated decision system means an automated decision system used by a utility that does any of the following:
(1)Creates or modifies system records, including geographic information system (GIS) layers, asset registers, and configuration logs, used in operational or engineering decisionmaking.
(2)Issues, prepopulates, or prioritizes operational actions, such as switching, fault isolation, circuit reconfiguration, load transfers, or gas-flow adjustments.
(3)Prioritizes safety critical or wildfire risk mitigation decisions, including, any of the following:
(A)Line deenergization or vegetation clearance.
(B)The classification or escalation of suspected gas leaks or electrical hazards.
(C)Recommendations affecting whether or when emergency response or field personnel dispatch is initiated.
(4)Generates, prioritizes, or recommends either of the following:
(A)Supervisory control and data acquisition.
(B)Control room or operational control actions that alter operating states, including set points, valve lineups, compressor controls, pressure or flow adjustments, alarms, interlocks, or operating limits.
(5)Creates, modifies, or recommends changes to safety critical system-of-record data that governs utility operations, including asset configuration data, isolation or valve lists, lockout or tagout points or procedures, alarm rationalization data, management of change records, or integrity-related operating parameters.
(g)Provenance log means a traceable record that identifies the artificial intelligence model used, its version, its training data source, the time the model output was generated, the identity of any human reviewer, and the actions taken that are approved, modified, or rejected by the human reviewer.
(h)Staging mode means the operating mode of a system where changes proposed by an artificial intelligence system are held in a nonoperational environment pending human review, testing, or validation before deployment into the live system.
(i)System records means design documents, GIS layers, configuration logs or files, change-management records, mapping files, asset registers, or other digital or analog records used by a covered utility to design, plan, configure, operate, maintain, or oversee electrical or gas systems.
(j)Technological change means the introduction of new or altered technology, equipment, software, automation, or robotics, or a new or altered artificial intelligence application, that alters the type, manner, or amount of work performed by employees of a covered utility.
(k)Utility mapping system means a GIS or other digital platform used by a covered utility to record, monitor, design, configure, or manage the layout, configuration, status, or changes of utility infrastructure circuits, pipelines, substations, valves, or other equipment or assets.
(l)Vendor automated decision system supply chain disclosure means the documentation provided by a vendor of an automated decision system that identifies model family, version, training data sources, update cadence, data residency, and known limitations or biases.
8511.The commission shall oversee the implementation of this chapter by a covered privately owned utility. The Energy Commission shall oversee the implementation of this chapter by a covered publicly owned utility. The commission and Energy Commission shall coordinate their actions pursuant to this chapter to ensure that the requirements for covered privately owned utilities and covered publicly owned utilities are consistent.
8512.(a)Every covered utility that employs an automated decision system in the mapping, design, configuration, operation, maintenance, or oversight of electrical or gas infrastructure shall maintain a structured process by which qualified personnel are able modify or override the output of the automated decision systems.
(b)A covered utility shall not implement the output of a high-risk automated decision system without prior affirmative human review and approval of that output.
(c)If the output of a high-risk automated decision system constitutes an engineering decision, the human review and approval required by subdivision (b) shall be performed by a California-licensed professional engineer. The high-risk automated decision system shall function as a decision-support tool and shall not independently execute operational actions or modify system records.
(d)The covered utility shall maintain a provenance log for all automated decision systems outputs and human review actions, including timestamps, model version, training data source, human approvers identity, modifications and overrides, and outcome of the review.
8513.(a)A covered utility shall not deploy a high-risk automated decision system in its live operational environment unless it first files with the commission or Energy Commission, as appropriate, a safety plan that includes all of the following:
(1)Information about the high-risk automated decision system that includes all of the following:
(A)The model version.
(B)Training data description.
(C)Update cadence.
(D)Known limitations.
(E)Bias testing.
(F)Cybersecurity controls.
(G)Human-override provisions.
(H)Roll-back features.
(I)Logging capability.
(2)Scenario testing results, including, at minimum, one wildfire or hazard scenario and one distributed energy resource-congestion or equipment-failure scenario, showing system behavior, failsafe behavior, human override, and system recovery.
(3)A plan for monitoring the performance, periodic validation, and retraining or decommissioning of high-risk automated decision systems if performance falls below safety thresholds.
(b)Before full deployment in a covered utilitys operation, the high-risk automated decision system shall operate in staging mode for a minimum of 18 months, or a commission-determined or Energy Commission-determined time period, as appropriate, with all changes audited and human approved before deployment to a live operational environment.
(c)A covered utility shall provide a copy of the safety plan required by subdivision (a) to affected labor organizations upon filing or within 15 days upon request.
8514.(a)A high-risk automated decision system that creates, modifies, updates, or purports to correct system records, including GIS layers, asset registers, and configuration logs, of a covered utility shall do all of the following:
(1)Operate initially in staging mode until human review is complete.
(2)Store proposed changes in a separate environment clearly labeled proposed by ADS pending human review and only be pushed to operational use after human approval.
(3)Maintain a provenance log capturing model, version, training data source, human reviewer identity, review timestamp, modification approval or rejection, and system version history.
(4)Include a roll-back mechanism enabling restoration of prior versions of system records in the event of a downstream adverse effect or error detection.
(b)The covered utility shall maintain audit trails of all changes stemming from automated decision systems and human review for a minimum of ____ years and make the audit trails available to the commission or Energy commission, as appropriate, upon request.
8515.(a)A covered utility shall report to the commission or Energy Commission, as appropriate, within 24 hours of discovering any event in which a high-risk automated decision system contributed to, or caused, any of the following:
(1)A service interruption or outage affecting more than 500 customers.
(2)Equipment damage, a failure, or a safety hazard.
(3)A misswitching event, incorrect mapping or asset register update, erroneous configuration change, or data-integrity breach with safety or reliability implications.
(b)Within 30 days of the occurrence of an event specified in subdivision (a), a root-cause report shall be submitted to the commission or Energy Commission, as appropriate, that includes both of the following:
(1)The role of the high-risk automated decision system, vendor model and version of the high-risk automated decision system used, training data source, human oversight steps, and provenance log.
(2)Corrective actions taken, scheduling of remediation, human-override review of the high-risk automated decision system, and actionable changes to safeguards.
(c)(1)A covered privately owned utility shall continuously monitor its high-risk automated decision systems and submit an annual report to the commission on its risk-assessment update, performance versus baseline, any human-override statistics, and any near-misses or incidents.
(2)A covered publicly owned utility shall continuously monitor its high-risk automated decision systems and submit an annual report to the Energy Commission and to its governing board on its risk-assessment update, performance versus baseline, any human-override statistics, and any near-misses or incidents.
8516.(a)A covered utility shall provide at least 180 days advance notice to affected labor organizations and impacted employee classifications before introducing any technological change involving automated decision systems that materially affects job duties, classifications, staffing levels, or required training.
(b)The notice shall include all of the following:
(1)A description of the proposed technology, its intended function, and a timeline for deployment.
(2)The anticipated or actual impacts on work processes, staffing, skill sets, and employee classifications.
(3)A plan for retraining, redeployment, or reassignment of affected employees.
(4)A demonstration that the proposed technology will not erode the work of licensed engineers, designers, or technical employees, unless retraining and deployment are offered.
(c)Covered utilities shall develop joint retraining programs, in consultation with employee representatives, to ensure employees can transition to new work roles created by the technological change.
(d)A covered utility shall not implement a high-risk automated decision system deployment in its operation that results in layoff of employees engaged in engineering decisionmaking, mapping, design, or technical operations unless the covered utility has first exhausted feasible retraining, redeployment, or reclassification options.
8517.The commission and Energy Commission may adopt regulations, guidelines, or procedures, as appropriate, to implement this chapter.
8518.(a)A covered privately owned utility that violates this chapter may be subject to enforcement by the commission under its authority over safe and reliable operations and is subject to penalties under Section 2108, Section 2110, or other applicable provisions of this code.
(b)A covered publicly owned utility shall annually certify to its governing board and Energy Commission that it is in compliance with this chapter and any adopted regulations, guidelines, or procedures implementing this chapter.
8519.(a)This chapter establishes minimum safety and governance standards of automated decision systems.
(b)This chapter does not limit, waive, or alter any rights, remedies, or obligations under state or federal law, including the National Labor Relations Act (29 U.S.C. Sec. 151 et seq.), the Meyers-Milias-Brown Act (Chapter 10 (commencing with Section 3500) of Division 4 of Title 1 of the Government Code), the Ralph C. Dills Act (Chapter 10.3 (commencing with Section 3512) of Division 4 of Title 1 of the Government Code), or any collective bargaining agreement, with respect to technological change, staffing, workload, training, or working conditions.
8520.The provisions of this chapter are severable. If any provision of this chapter or its application is held invalid, that invalidity shall not affect other provisions or applications that can be given effect without the invalid provision or application.
SEC. 4.No reimbursement is required by this act pursuant to Section 6 of Article XIIIB of the California Constitution because a local agency or school district has the authority to levy service charges, fees, or assessments sufficient to pay for the program or level of service mandated by this act or because costs that may be incurred by a local agency or school district will be incurred because this act creates a new crime or infraction, eliminates a crime or infraction, or changes the penalty for a crime or infraction, within the meaning of Section 17556 of the Government Code, or changes the definition of a crime within the meaning of Section 6 of Article XIIIB of the California Constitution.
|
Date
|
Status
|
|---|---|
| May 28, 2026 | Ordered to inactive file on request of Senator McNerney. |
| May 19, 2026 | Read second time. Ordered to third reading. |
| May 18, 2026 | From committee: Do pass as amended. (Ayes 5. Noes 2. Page 4261.) (May 14). |
| May 18, 2026 | Read second time and amended. Ordered to second reading. |
| May 18, 2026 | From committee: Do pass as amended. (Ayes 5. Noes 2.) (May 14). |
| May 8, 2026 | Set for hearing May 14. |
| May 4, 2026 | May 4 hearing: Placed on APPR. suspense file. |
| Apr 24, 2026 | Set for hearing May 4. |
| Apr 22, 2026 | Read second time and amended. Re-referred to Com. on APPR. |
| Apr 21, 2026 | From committee: Do pass as amended and re-refer to Com. on APPR. (Ayes 7. Noes 2. Page 3957.) (April 20). |
| Apr 21, 2026 | From committee: Do pass as amended and re-refer to Com. on APPR. (Ayes 7. Noes 2.) (April 20). |
| Apr 14, 2026 | From committee: Do pass and re-refer to Com. on P., D.T., & C.P. (Ayes 13. Noes 3. Page 3842.) (April 13). Re-referred to Com. on P., D.T., & C.P. |
| Apr 14, 2026 | From committee: Do pass and re-refer to Com. on P., D.T., & C.P. (Ayes 13. Noes 3.) (April 13). Re-referred to Com. on P., D.T., & C.P. |
| Apr 10, 2026 | Set for hearing April 20 in P., D.T., & C.P. pending receipt. |
| Apr 2, 2026 | Set for hearing April 13. |
| Feb 18, 2026 | Referred to Coms. on E., U & C. and P., D.T., & C.P. |
| Feb 11, 2026 | From printer. May be acted upon on or after March 13. |
| Feb 10, 2026 | Introduced. Read first time. To Com. on RLS. for assignment. To print. |
|
Date
|
Description
|
Location
|
Votes
|
|
|---|---|---|---|---|
| May 14, 2026 | Do pass as amended | Senate Standing Committee on Appropriations |
Passed |
|
| May 4, 2026 | Placed on suspense file | Senate Standing Committee on Appropriations |
Passed |
|
| Apr 20, 2026 | Do pass as amended, but first amend, and re-refer to the Committee on [Appropriations] | Senate Standing Committee on Privacy, Digital Technologies, and Consumer Protection |
Passed |
|
| Apr 13, 2026 | Do pass, but first be re-referred to the Committee on [Privacy, Digital Technologies, and Consumer Protection] | Senate Standing Committee on Energy, Utilities and Communications |
Passed |
This list shows positions taken on this bill by organizations on the date displayed, which is linked to the original source.
Positions by individuals are not included.
|
Organization
|
Position
|
Date
|
|---|---|---|
| Board for Professional Engineers, Land Surveyors, and Geologists | ** Against | |
| Burbank Water and Power | ** Against | |
| California Alliance For Retired Americans | ** Support | |
| California Chamber Of Commerce | ** Against | |
| California Federation Labor Unions, AFL-CIO | ** Support | |
| California Labor Federation, Afl-Cio | ** Support | |
| California Municipal Utilities Association | ** Against | |
| California School Employees Association, AFL-CIO | ** Support | |
| Engineers and Scientists of California, Local 20, IFPTE | ** Support | |
| Imperial Irrigation District | ** Against | |
| Oakland Privacy | ** Support | |
| San Diego Gas and Electric Company | ** Against | |
| Southern California Gas Company | ** Against | |
| Southern California Public Power Authority | ** Against | |
| Utility Workers Union Of America, Local 132 | ** Support | |
| Utility Workers Union of America, Local 483 | ** Support | |
| Utility Workers Union of America, Local 522 | ** Support | |
| Utility Workers Union Of America, Local 522 | ** Support | |
| Utility Workers Union of America Locals 132, 483, and 522 | ** Support |
SENATE RULES COMMITTEE Office of Senate Floor Analyses (916) 651-4171 | SB 1011 |
THIRD READING
Bill No: SB 1011
Author: McNerney (D)
Amended: 5/18/26
Vote: 21
SENATE ENERGY, U. & C. COMMITTEE: 13-3, 4/13/26
AYES: Allen, Archuleta, Arreguín, Becker, Caballero, Gonzalez, Hurtado, McNerney, Reyes, Richardson, Rubio, Stern, Wahab
NOES: Ochoa Bogh, Grove, Strickland
NO VOTE RECORDED: Dahle
SENATE PRIV., DIGITAL TECH. & CONS. PROT. COMMITTEE: 7-2, 4/20/26
AYES: Cabaldon, Gonzalez, McNerney, Padilla, Reyes, Umberg, Wiener
NOES: Jones, Seyarto
SENATE APPROPRIATIONS COMMITTEE: 5-2, 5/14/26
AYES: Cervantes, Cabaldon, Grayson, Richardson, Wahab
NOES: Seyarto, Dahle
SUBJECT: Energy: Utility Infrastructure AI Safety, Oversight, and Workforce Protection Act
SOURCE: California Federation Labor Unions, AFL-CIO
California School Employees Association, AFL-CIO
Engineers and Scientists of California, Local 20, IFPTE
Utility Workers Union of America, Local 132
Utility Workers Union of America, Local 483
Utility Workers Union of America, Local 522
DIGEST:
This bill (1) requires the California Public Utilities Commission (CPUC) to adopt standards by January 1, 2020, for electrical and gas corporations’ use of artificial intelligence (AI) models; (2) specifies certain requirements that must be included in the CPUC’s standards for utility AI models, including but not limited to, requirements for human review of AI models used in certain utility operations; (3) requires the CPUC to direct electrical and gas corporations to file plans with the CPUC demonstrating compliance with the CPUC’s AI standards and authorizes the CPUC to prohibit a gas or electrical corporation’s use of an AI model if the CPUC finds that the model’s use would negatively impact the safe, reliable and affordable provision of utility services (4) requires the boards of Community Choice Aggregators (CCAs) and local publicly owned electric utilities (POUs) to adopt policies regarding their use of AI models that are consistent with standards adopted by the CPUC.
ANALYSIS:
Existing law:
- Establishes the CPUC, consisting of five members appointed by the Governor, and authorizes the CPUC to fix rates and establish rules for public utilities. (Article XII of the California Constitution)
- Requires the CPUC to establish rules for a public utility and revise those requirements through an order or rule whenever the CPUC, after a hearing, finds that existing rules, practices, equipment, appliances, facilities or service of any public utility or manufacturing, distribution, transmission, storage or supply methods employed by the public utility are unjust, unreasonable, unsafe, improper, inadequate, or insufficient. (Public Utilities Code §761)
- Requires every public utility to establish and maintain adequate, efficient, just, and reasonable services, instruments, equipment and facilities needed to promote the health, safety and comfort of customers, employees, and the public. Any rule adopted by a public utility regarding its services or rates for the public must be just and reasonable. (Public Utilities Code §451)
- Establishes the California Wildfire Safety Advisory Board, consisting of seven members to advise utilities and the Office of Energy Infrastructure Safety (OEIS) on utility wildfire safety issues, as specified. (Public Utilities Code §326.1)
- Establishes the OEIS in the Natural Resources Agency to oversee electrical corporations’ compliance with wildfire safety requirements, as specified. (Government Code §15470 et. seq.)
- Establishes the CEC, consisting of five members appointed by the Governor, and specifies the duties of the CEC. Every two years, the Governor must designate a chair and vice chair from the CEC’s membership. The CEC must appoint a public adviser every three years to carry out certain public engagement duties. (Public Resources Code §25200 et. seq.)
- Defines an “artificial intelligence model” as an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments. Existing law establishes requirements for large developers of artificial intelligence models. (Business and Professions Code §22757.10 et. seq.)
This bill:
- Defines an “artificial intelligence model” as an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments.
- Requires the CPUC to adopt standards by January 1, 2028, for an electrical or gas corporation’s use of AI models. This bill specifies that the CPUC’s AI standards must do at least the following:
- Establish disclosure requirements for an electrical corporation or gas corporation to specify the types of AI models it uses and how those models are used by the electrical corporation or gas corporation.
- Require an electrical corporation or gas corporation to identify the employee job classifications impacted by any planned implementation of an AI model and efforts taken to ensure effective education, training, and retention for impacted employees.
- Identify steps an electrical corporation or gas corporation shall take to ensure that AI models do not impact utility safety, affordability, and reliability.
- Establish requirements for human review and approval for the deployment and use of AI models to ensure that electrical and gas services are safe, affordable, and reliable, including requirements for the placement of human review in the deployment of AI models.
- Ensure that deployment of artificial intelligence models do not displace utility employees needed for the safe, affordable, and reliable provision of electrical and gas services, as specified.
- Establish requirements for human review and approval of any artificial intelligence that is high risk, safety sensitive, or would have a material operational impact.
- Authorizes the CPUC to prohibit an electrical or gas corporation’s use of an AI model if the commission finds that deployment of the model would negatively impact the provision of safe, affordable, and reliable electrical or gas service.
- Requires the CPUC to direct an electrical or gas corporation to file a plan demonstrating compliance with the CPUC’s AI model standards. The CPUC may request records it deems necessary to verify compliance with its standards.
- Requires the CPUC to ensure that OEIS can review any electrical corporation’s AI model plan to ensure that an AI model used for wildfire mitigation purposes is consistent with the electrical corporation’s wildfire mitigation plan.
- Require the CPUC to ensure that impacted bargaining units are consulted in the development plans filed at the CPUC regarding utilities’ use of AI models.
- Requires each CCA and POU to adopt a policy regarding their respective use of AI models. This bill requires CCA and POU policies to be consistent with the standards adopted by the CPUC.
Background
How do utilities use AI? Utilities’ use of generative AI is nascent; however, some utilities are seeking ways to develop AI tools that are trained to help with specific power sector needs. Diablo Canyon Power Plant partnered with Atomic Canyon to deploy a generative AI system that uses the Nuclear Regulatory Commission database and the Oak Ridge National Laboratory’s Frontier supercomputer to search data from Diablo Canyon’s files to synthesize information for federal regulatory filings. This use of AI is the first commercial use of generative AI at any nuclear power facility in the nation. The use of AI can have benefits; however, some recent demonstrations of AI have also shown that generative AI can also lead to unsafe and erroneous decisions. Several California gas and electric utilities are starting to integrate AI into efforts to prevent and respond to wildfires. Electric utilities have turned to AI solutions to obtain better data identifying infrastructure with greater risks for igniting wildfires and use of cameras with AI data analysis to identify potential fires before they grow. Early results from the use of these AI wildfire mitigation tools indicates that the tools are successful at reducing wildfire ignitions associated with utility infrastructure and provide early notifications to emergency personnel when fires are spotted.
Multiple agencies oversee utility safety requirements, but few standards exist for AI utility use. While AI can provide substantial resources to improve utility safety and reliability, AI can also pose risks to the energy sector. Biases in algorithms or poor AI training can lead to systemically reproduced biased and erroneous outcomes. Superintelligent AI systems are capable of creating systems that lack sufficient human input to stop harmful outcomes. The CPUC is one of several state and federal agencies that have adopted rules setting forth specific requirements for certain utility safety measures. Multiple CPUC general orders establish requirements for safe installation and operation of electric and natural gas infrastructure. Under existing utility safety rules, utilities under the CPUC’s jurisdiction must report certain critical incidents to the CPUC within two working hours or four non-working hours if the incident resulted in death, hospitalization, major news coverage, significant property damage, or prompted an investigation regarding potential utility wildfire ignition. Additionally, CPUC General Order 166 (GO 166) establishes standards for electric utilities’ operation during emergencies and disasters, which includes mandatory reporting regarding major power outages. While the CPUC has adopted specific requirements regarding safety measures for electrical and gas infrastructure, the CPUC has not adopted specific safety requirements for utilities’ use of AI when operating electrical and gas infrastructure. The CPUC generally has broad authority to regulate utilities to ensure the provision of safe and reliable gas and electric service at just and reasonable rates; however, it is unclear if CPUC rules have been sufficiently reviewed and updated to reflect the use of AI in utility operations. It is also not clear whether the CPUC has information from all the utilities regarding the types of AI models used. While some AI models may pose limited risks, other AI models that use more intelligent generative AI functions can pose greater risks by amplifying biases and creating data “hallucinations.” This bill would expand disclosures of electrical and gas utilities’ use of AI models and expand the CPUC’s authority to regulate these utilities’ use of AI.
Bill requires the CPUC to regulate human review of utilities’ AI operations. This bill requires the CPUC to establish standards for utilities’ AI models, and this bill requires those standards to include specific provisions regarding human review of AI processes in utility operations. Specifically, this bill requires the CPUC to establish requirements for human review during the deployment of AI models. AI models may differ widely from each other and may be developed for highly customized purposes. To effectively evaluate the appropriate level of human review for each use of AI models in the energy utility sector, the CPUC may need to evaluate AI deployment usage on a case-by-case basis to ensure that AI used for utility safety and reliability is not impeded while preventing harmful AI outcomes.
Related/Prior Legislation
SB 947 (McNerney) of 2026, limits employers’ use of ADS in certain employment decisions and requires the Labor Commissioner to oversee employers’ use of ADS in workplace decisions. The bill is pending in the Senate.
SB 833 (McNerney) of 2025, establishes requirements for human oversight over AI systems used with critical infrastructure. The bill also defined AI as including ADS. The bill is pending in the Assembly.
SB 420 (Padilla) of 2025, regulates the use of high-risk ADS, including requiring ADS developers and deployers to performing impact assessments on their systems. The bill establishes the right of individuals to know when an ADS has been used, details about the systems, and an opportunity to appeal ADS decisions, where technically feasible. The bill is pending in the Assembly.
SB 53 (Weiner, Chapter 138, Statutes of 2025) required large developers of certain AI tools to adopt protocols to address catastrophic risks associated with AI, disclose certain information about their AI models, and submit specified reports to Office of Emergency Services.
SB 7 (McNerney) of 2025, would have established limitations and notices for the use of ADS in employment decisions. The bill was vetoed.
AB 1018 (Bauer-Kahan) of 2025, would have regulated the use of ADS. It established obligations on developers and deployers of ADS designed or used to make or facilitate “consequential decisions. The bill is on the Senate Inactive File.
FISCAL EFFECT: Appropriation: No Fiscal Com.: Yes Local: Yes
According to the Senate Appropriations Committee, ongoing costs likely in the high hundreds of thousands of dollars annually (ratepayer funds) for the CPUC to adopt standards governing the use of AI models in utility operations, review and evaluate utility AI safety plans for compliance, establish oversight and enforcement mechanisms to ensure adherence to adopted standards, and assess and monitor any impacts to the provision of safe, affordable, and reliable electrical or gas service.
SUPPORT: (Verified 5/14/26)
California Federation Labor Unions, AFL-CIO (Co-source)
California School Employees Association, AFL-CIO (Co-source)
Engineers and Scientists of California, Local 20, IFPTE (Co-source)
Utility Workers Union of America, Local 132 (Co-source)
Utility Workers Union of America, Local 483 (Co-source)
Utility Workers Union of America, Local 522 (Co-source)
California Alliance for Retired Americans
Oakland Privacy
OPPOSITION: (Verified 5/14/26)
Board for Professional Engineers, Land Surveyors, and Geologists
Burbank Water and Power
California Chamber of Commerce
California Municipal Utilities Association
Imperial Irrigation District
San Diego Gas and Electric Company
Southern California Gas Company
Southern California Public Power Authority
ARGUMENTS IN SUPPORT: According to the author:
AI has the potential to improve efficiency in California’s power grid and quicken response to wildfires and life-threatening dangers, but AI also presents significant risks. AI can hallucinate – make mistakes that could pose serious threats. Without human oversight, unchecked AI could trigger catastrophic equipment failures, service outages for thousands of residents, or life-threatening safety hazards. California can’t afford to outsource the safety of our power grid to unproven, unmonitored algorithms. California also needs to protect workers in the power sector from widespread job displacement. SB 1011 is a commonsense solution that will reduce risk by ensuring that AI use supports human decision-making rather than replacing it.
ARGUMENTS IN OPPOSITION: Opponents argue that the CPUC is not equipped to regulate AI models and regulations adopted by the CPUC may impede the use of AI models that are beneficial in the utility sector. In opposition, the California Chamber of Commerce states:
The bill’s definitions of “artificial intelligence model” is so broad that the bill likely sweeps in a wide range of longstanding utility systems, including distribution and outage management systems, powerplant control systems, and platforms used to manage distributed energy resources. In doing so, the bill risks inadvertently capturing decades of standardized utility technology already embedded in day-to-day grid and natural gas system operations.
The Southern California Public Power Authority (SCPPA) opposes this bill, arguing that requiring POUs to adopt policies similar to those imposed by the CPUC on IOUs would extend the CPUC’s regulatory powers to locally-governed POUs. In opposition, SCPPA states:
It should also be noted that SB 1011 requires all POUs to develop these CPUC policies, regardless of whether the utility is actually using AI models or already has an existing policy governing its use of AI models. In these cases, the bill is an unnecessary or duplicative paper exercise that adds an undue administrative burden on not-for-profit POUs that operate on a local government budget. The utilities represented by SCPPA are public agencies, operated by public employees, and governed by locally elected officials. Utility operational decisions are guided by what is in the best interest of the public they serve. This includes reliability, safety and cost.
Prepared by: Sarah Smith / E., U. & C. / (916) 651-4107
5/20/26 13:33:47
**** END ****
SENATE RULES COMMITTEE Office of Senate Floor Analyses (916) 651-4171 | SB 1011 |
THIRD READING
Bill No: SB 1011
Author: McNerney (D)
Amended: 5/18/26
Vote: 21
SENATE ENERGY, U. & C. COMMITTEE: 13-3, 4/13/26
AYES: Allen, Archuleta, Arreguín, Becker, Caballero, Gonzalez, Hurtado, McNerney, Reyes, Richardson, Rubio, Stern, Wahab
NOES: Ochoa Bogh, Grove, Strickland
NO VOTE RECORDED: Dahle
SENATE PRIV., DIGITAL TECH. & CONS. PROT. COMMITTEE: 7-2, 4/20/26
AYES: Cabaldon, Gonzalez, McNerney, Padilla, Reyes, Umberg, Wiener
NOES: Jones, Seyarto
SENATE APPROPRIATIONS COMMITTEE: 5-2, 5/14/26
AYES: Cervantes, Cabaldon, Grayson, Richardson, Wahab
NOES: Seyarto, Dahle
SUBJECT: Energy: Utility Infrastructure AI Safety, Oversight, and Workforce Protection Act
SOURCE: California Federation Labor Unions, AFL-CIO
California School Employees Association, AFL-CIO
Engineers and Scientists of California, Local 20, IFPTE
Utility Workers Union of America, Local 132
Utility Workers Union of America, Local 483
Utility Workers Union of America, Local 522
DIGEST:
This bill (1) requires the California Public Utilities Commission (CPUC) to adopt standards by January 1, 2020, for electrical and gas corporations’ use of artificial intelligence (AI) models; (2) specifies certain requirements that must be included in the CPUC’s standards for utility AI models, including but not limited to, requirements for human review of AI models used in certain utility operations; (3) requires the CPUC to direct electrical and gas corporations to file plans with the CPUC demonstrating compliance with the CPUC’s AI standards and authorizes the CPUC to prohibit a gas or electrical corporation’s use of an AI model if the CPUC finds that the model’s use would negatively impact the safe, reliable and affordable provision of utility services (4) requires the boards of Community Choice Aggregators (CCAs) and local publicly owned electric utilities (POUs) to adopt policies regarding their use of AI models that are consistent with standards adopted by the CPUC.
ANALYSIS:
Existing law:
- Establishes the CPUC, consisting of five members appointed by the Governor, and authorizes the CPUC to fix rates and establish rules for public utilities. (Article XII of the California Constitution)
- Requires the CPUC to establish rules for a public utility and revise those requirements through an order or rule whenever the CPUC, after a hearing, finds that existing rules, practices, equipment, appliances, facilities or service of any public utility or manufacturing, distribution, transmission, storage or supply methods employed by the public utility are unjust, unreasonable, unsafe, improper, inadequate, or insufficient. (Public Utilities Code §761)
- Requires every public utility to establish and maintain adequate, efficient, just, and reasonable services, instruments, equipment and facilities needed to promote the health, safety and comfort of customers, employees, and the public. Any rule adopted by a public utility regarding its services or rates for the public must be just and reasonable. (Public Utilities Code §451)
- Establishes the California Wildfire Safety Advisory Board, consisting of seven members to advise utilities and the Office of Energy Infrastructure Safety (OEIS) on utility wildfire safety issues, as specified. (Public Utilities Code §326.1)
- Establishes the OEIS in the Natural Resources Agency to oversee electrical corporations’ compliance with wildfire safety requirements, as specified. (Government Code §15470 et. seq.)
- Establishes the CEC, consisting of five members appointed by the Governor, and specifies the duties of the CEC. Every two years, the Governor must designate a chair and vice chair from the CEC’s membership. The CEC must appoint a public adviser every three years to carry out certain public engagement duties. (Public Resources Code §25200 et. seq.)
- Defines an “artificial intelligence model” as an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments. Existing law establishes requirements for large developers of artificial intelligence models. (Business and Professions Code §22757.10 et. seq.)
This bill:
- Defines an “artificial intelligence model” as an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments.
- Requires the CPUC to adopt standards by January 1, 2028, for an electrical or gas corporation’s use of AI models. This bill specifies that the CPUC’s AI standards must do at least the following:
- Establish disclosure requirements for an electrical corporation or gas corporation to specify the types of AI models it uses and how those models are used by the electrical corporation or gas corporation.
- Require an electrical corporation or gas corporation to identify the employee job classifications impacted by any planned implementation of an AI model and efforts taken to ensure effective education, training, and retention for impacted employees.
- Identify steps an electrical corporation or gas corporation shall take to ensure that AI models do not impact utility safety, affordability, and reliability.
- Establish requirements for human review and approval for the deployment and use of AI models to ensure that electrical and gas services are safe, affordable, and reliable, including requirements for the placement of human review in the deployment of AI models.
- Ensure that deployment of artificial intelligence models do not displace utility employees needed for the safe, affordable, and reliable provision of electrical and gas services, as specified.
- Establish requirements for human review and approval of any artificial intelligence that is high risk, safety sensitive, or would have a material operational impact.
- Authorizes the CPUC to prohibit an electrical or gas corporation’s use of an AI model if the commission finds that deployment of the model would negatively impact the provision of safe, affordable, and reliable electrical or gas service.
- Requires the CPUC to direct an electrical or gas corporation to file a plan demonstrating compliance with the CPUC’s AI model standards. The CPUC may request records it deems necessary to verify compliance with its standards.
- Requires the CPUC to ensure that OEIS can review any electrical corporation’s AI model plan to ensure that an AI model used for wildfire mitigation purposes is consistent with the electrical corporation’s wildfire mitigation plan.
- Require the CPUC to ensure that impacted bargaining units are consulted in the development plans filed at the CPUC regarding utilities’ use of AI models.
- Requires each CCA and POU to adopt a policy regarding their respective use of AI models. This bill requires CCA and POU policies to be consistent with the standards adopted by the CPUC.
Background
How do utilities use AI? Utilities’ use of generative AI is nascent; however, some utilities are seeking ways to develop AI tools that are trained to help with specific power sector needs. Diablo Canyon Power Plant partnered with Atomic Canyon to deploy a generative AI system that uses the Nuclear Regulatory Commission database and the Oak Ridge National Laboratory’s Frontier supercomputer to search data from Diablo Canyon’s files to synthesize information for federal regulatory filings. This use of AI is the first commercial use of generative AI at any nuclear power facility in the nation. The use of AI can have benefits; however, some recent demonstrations of AI have also shown that generative AI can also lead to unsafe and erroneous decisions. Several California gas and electric utilities are starting to integrate AI into efforts to prevent and respond to wildfires. Electric utilities have turned to AI solutions to obtain better data identifying infrastructure with greater risks for igniting wildfires and use of cameras with AI data analysis to identify potential fires before they grow. Early results from the use of these AI wildfire mitigation tools indicates that the tools are successful at reducing wildfire ignitions associated with utility infrastructure and provide early notifications to emergency personnel when fires are spotted.
Multiple agencies oversee utility safety requirements, but few standards exist for AI utility use. While AI can provide substantial resources to improve utility safety and reliability, AI can also pose risks to the energy sector. Biases in algorithms or poor AI training can lead to systemically reproduced biased and erroneous outcomes. Superintelligent AI systems are capable of creating systems that lack sufficient human input to stop harmful outcomes. The CPUC is one of several state and federal agencies that have adopted rules setting forth specific requirements for certain utility safety measures. Multiple CPUC general orders establish requirements for safe installation and operation of electric and natural gas infrastructure. Under existing utility safety rules, utilities under the CPUC’s jurisdiction must report certain critical incidents to the CPUC within two working hours or four non-working hours if the incident resulted in death, hospitalization, major news coverage, significant property damage, or prompted an investigation regarding potential utility wildfire ignition. Additionally, CPUC General Order 166 (GO 166) establishes standards for electric utilities’ operation during emergencies and disasters, which includes mandatory reporting regarding major power outages. While the CPUC has adopted specific requirements regarding safety measures for electrical and gas infrastructure, the CPUC has not adopted specific safety requirements for utilities’ use of AI when operating electrical and gas infrastructure. The CPUC generally has broad authority to regulate utilities to ensure the provision of safe and reliable gas and electric service at just and reasonable rates; however, it is unclear if CPUC rules have been sufficiently reviewed and updated to reflect the use of AI in utility operations. It is also not clear whether the CPUC has information from all the utilities regarding the types of AI models used. While some AI models may pose limited risks, other AI models that use more intelligent generative AI functions can pose greater risks by amplifying biases and creating data “hallucinations.” This bill would expand disclosures of electrical and gas utilities’ use of AI models and expand the CPUC’s authority to regulate these utilities’ use of AI.
Bill requires the CPUC to regulate human review of utilities’ AI operations. This bill requires the CPUC to establish standards for utilities’ AI models, and this bill requires those standards to include specific provisions regarding human review of AI processes in utility operations. Specifically, this bill requires the CPUC to establish requirements for human review during the deployment of AI models. AI models may differ widely from each other and may be developed for highly customized purposes. To effectively evaluate the appropriate level of human review for each use of AI models in the energy utility sector, the CPUC may need to evaluate AI deployment usage on a case-by-case basis to ensure that AI used for utility safety and reliability is not impeded while preventing harmful AI outcomes.
Related/Prior Legislation
SB 947 (McNerney) of 2026, limits employers’ use of ADS in certain employment decisions and requires the Labor Commissioner to oversee employers’ use of ADS in workplace decisions. The bill is pending in the Senate.
SB 833 (McNerney) of 2025, establishes requirements for human oversight over AI systems used with critical infrastructure. The bill also defined AI as including ADS. The bill is pending in the Assembly.
SB 420 (Padilla) of 2025, regulates the use of high-risk ADS, including requiring ADS developers and deployers to performing impact assessments on their systems. The bill establishes the right of individuals to know when an ADS has been used, details about the systems, and an opportunity to appeal ADS decisions, where technically feasible. The bill is pending in the Assembly.
SB 53 (Weiner, Chapter 138, Statutes of 2025) required large developers of certain AI tools to adopt protocols to address catastrophic risks associated with AI, disclose certain information about their AI models, and submit specified reports to Office of Emergency Services.
SB 7 (McNerney) of 2025, would have established limitations and notices for the use of ADS in employment decisions. The bill was vetoed.
AB 1018 (Bauer-Kahan) of 2025, would have regulated the use of ADS. It established obligations on developers and deployers of ADS designed or used to make or facilitate “consequential decisions. The bill is on the Senate Inactive File.
FISCAL EFFECT: Appropriation: No Fiscal Com.: Yes Local: Yes
According to the Senate Appropriations Committee, ongoing costs likely in the high hundreds of thousands of dollars annually (ratepayer funds) for the CPUC to adopt standards governing the use of AI models in utility operations, review and evaluate utility AI safety plans for compliance, establish oversight and enforcement mechanisms to ensure adherence to adopted standards, and assess and monitor any impacts to the provision of safe, affordable, and reliable electrical or gas service.
SUPPORT: (Verified 5/14/26)
California Federation Labor Unions, AFL-CIO (Co-source)
California School Employees Association, AFL-CIO (Co-source)
Engineers and Scientists of California, Local 20, IFPTE (Co-source)
Utility Workers Union of America, Local 132 (Co-source)
Utility Workers Union of America, Local 483 (Co-source)
Utility Workers Union of America, Local 522 (Co-source)
California Alliance for Retired Americans
Oakland Privacy
OPPOSITION: (Verified 5/14/26)
Burbank Water and Power
California Chamber of Commerce
California Municipal Utilities Association
Imperial Irrigation District
San Diego Gas and Electric Company
Southern California Gas Company
Southern California Public Power Authority
ARGUMENTS IN SUPPORT: According to the author:
AI has the potential to improve efficiency in California’s power grid and quicken response to wildfires and life-threatening dangers, but AI also presents significant risks. AI can hallucinate – make mistakes that could pose serious threats. Without human oversight, unchecked AI could trigger catastrophic equipment failures, service outages for thousands of residents, or life-threatening safety hazards. California can’t afford to outsource the safety of our power grid to unproven, unmonitored algorithms. California also needs to protect workers in the power sector from widespread job displacement. SB 1011 is a commonsense solution that will reduce risk by ensuring that AI use supports human decision-making rather than replacing it.
ARGUMENTS IN OPPOSITION: Opponents argue that the CPUC is not equipped to regulate AI models and regulations adopted by the CPUC may impede the use of AI models that are beneficial in the utility sector. In opposition, the California Chamber of Commerce states:
The bill’s definitions of “artificial intelligence model” is so broad that the bill likely sweeps in a wide range of longstanding utility systems, including distribution and outage management systems, powerplant control systems, and platforms used to manage distributed energy resources. In doing so, the bill risks inadvertently capturing decades of standardized utility technology already embedded in day-to-day grid and natural gas system operations.
The Southern California Public Power Authority (SCPPA) opposes this bill, arguing that requiring POUs to adopt policies similar to those imposed by the CPUC on IOUs would extend the CPUC’s regulatory powers to locally-governed POUs. In opposition, SCPPA states:
It should also be noted that SB 1011 requires all POUs to develop these CPUC policies, regardless of whether the utility is actually using AI models or already has an existing policy governing its use of AI models. In these cases, the bill is an unnecessary or duplicative paper exercise that adds an undue administrative burden on not-for-profit POUs that operate on a local government budget. The utilities represented by SCPPA are public agencies, operated by public employees, and governed by locally elected officials. Utility operational decisions are guided by what is in the best interest of the public they serve. This includes reliability, safety and cost.
Prepared by: Sarah Smith / E., U. & C. / (916) 651-4107
5/19/26 12:47:55
**** END ****
SENATE RULES COMMITTEE Office of Senate Floor Analyses (916) 651-4171 | SB 1011 |
THIRD READING
Bill No: SB 1011
Author: McNerney (D)
Amended: 4/22/26
Vote: 21
SENATE ENERGY, U. & C. COMMITTEE: 13-3, 4/13/26
AYES: Allen, Archuleta, Arreguín, Becker, Caballero, Gonzalez, Hurtado, McNerney, Reyes, Richardson, Rubio, Stern, Wahab
NOES: Ochoa Bogh, Grove, Strickland
NO VOTE RECORDED: Dahle
SENATE PRIV., DIGITAL TECH. & CONS. PROT. COMMITTEE: 7-2, 4/20/26
AYES: Cabaldon, Gonzalez, McNerney, Padilla, Reyes, Umberg, Wiener
NOES: Jones, Seyarto
SENATE APPROPRIATIONS COMMITTEE: 5-2, 5/14/26
AYES: Cervantes, Cabaldon, Grayson, Richardson, Wahab
NOES: Seyarto, Dahle
SUBJECT: Energy: Utility Infrastructure AI Safety, Oversight, and Workforce Protection Act
SOURCE: California Federation Labor Unions, AFL-CIO
California School Employees Association, AFL-CIO
Engineers and Scientists of California, Local 20, IFPTE
Utility Workers Union of America, Local 132
Utility Workers Union of America, Local 483
Utility Workers Union of America, Local 522
DIGEST:
This bill (1) requires the California Public Utilities Commission (CPUC) to adopt standards by January 1, 2020, for electrical and gas corporations’ use of artificial intelligence (AI) models; (2) specifies certain requirements that must be included in the CPUC’s standards for utility AI models, including but not limited to, requirements for human review of AI models used in certain utility operations; (3) requires the CPUC to direct electrical and gas corporations to file plans with the CPUC demonstrating compliance with the CPUC’s AI standards and authorizes the CPUC to prohibit a gas or electrical corporation’s use of an AI model if the CPUC finds that the model’s use would negatively impact the safe, reliable and affordable provision of utility services (4) requires the boards of Community Choice Aggregators (CCAs) and local publicly owned electric utilities (POUs) to adopt policies regarding their use of AI models that are consistent with standards adopted by the CPUC.
ANALYSIS:
Existing law:
- Establishes the CPUC, consisting of five members appointed by the Governor, and authorizes the CPUC to fix rates and establish rules for public utilities. (Article XII of the California Constitution)
- Requires the CPUC to establish rules for a public utility and revise those requirements through an order or rule whenever the CPUC, after a hearing, finds that existing rules, practices, equipment, appliances, facilities or service of any public utility or manufacturing, distribution, transmission, storage or supply methods employed by the public utility are unjust, unreasonable, unsafe, improper, inadequate, or insufficient. (Public Utilities Code §761)
- Requires every public utility to establish and maintain adequate, efficient, just, and reasonable services, instruments, equipment and facilities needed to promote the health, safety and comfort of customers, employees, and the public. Any rule adopted by a public utility regarding its services or rates for the public must be just and reasonable. (Public Utilities Code §451)
- Establishes the California Wildfire Safety Advisory Board, consisting of seven members to advise utilities and the Office of Energy Infrastructure Safety (OEIS) on utility wildfire safety issues, as specified. (Public Utilities Code §326.1)
- Establishes the OEIS in the Natural Resources Agency to oversee electrical corporations’ compliance with wildfire safety requirements, as specified. (Government Code §15470 et. seq.)
- Establishes the CEC, consisting of five members appointed by the Governor, and specifies the duties of the CEC. Every two years, the Governor must designate a chair and vice chair from the CEC’s membership. The CEC must appoint a public adviser every three years to carry out certain public engagement duties. (Public Resources Code §25200 et. seq.)
- Defines an “artificial intelligence model” as an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments. Existing law establishes requirements for large developers of artificial intelligence models. (Business and Professions Code §22757.10 et. seq.)
This bill:
- Defines an “artificial intelligence model” as an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments.
- Requires the CPUC to adopt standards by January 1, 2028, for an electrical or gas corporation’s use of AI models. This bill specifies that the CPUC’s AI standards must do at least the following:
- Establish disclosure requirements for an electrical corporation or gas corporation to specify the types of AI models it uses and how those models are used by the electrical corporation or gas corporation.
- Require an electrical corporation or gas corporation to identify the employee job classifications impacted by any planned implementation of an AI model and efforts taken to ensure effective education, training, and retention for impacted employees.
- Identify steps an electrical corporation or gas corporation shall take to ensure that AI models do not impact utility safety, affordability, and reliability.
- Establish requirements for human review and approval for the deployment and use of AI models to ensure that electrical and gas services are safe, affordable, and reliable, including requirements for the placement of human review in the deployment of AI models.
- Ensure that deployment of artificial intelligence models do not displace utility employees needed for the safe, affordable, and reliable provision of electrical and gas services, as specified.
- Establish requirements for human review and approval of any artificial intelligence that is high risk, safety sensitive, or would have a material operational impact.
- Authorizes the CPUC to prohibit an electrical or gas corporation’s use of an AI model if the commission finds that deployment of the model would negatively impact the provision of safe, affordable, and reliable electrical or gas service.
- Requires the CPUC to direct an electrical or gas corporation to file a plan demonstrating compliance with the CPUC’s AI model standards. The CPUC may request records it deems necessary to verify compliance with its standards.
- Requires the CPUC to ensure that OEIS can review any electrical corporation’s AI model plan to ensure that an AI model used for wildfire mitigation purposes is consistent with the electrical corporation’s wildfire mitigation plan.
- Require the CPUC to ensure that impacted bargaining units are consulted in the development plans filed at the CPUC regarding utilities’ use of AI models.
- Requires each CCA and POU to adopt a policy regarding their respective use of AI models. This bill requires CCA and POU policies to be consistent with the standards adopted by the CPUC.
Background
How do utilities use AI? Utilities’ use of generative AI is nascent; however, some utilities are seeking ways to develop AI tools that are trained to help with specific power sector needs. Diablo Canyon Power Plant partnered with Atomic Canyon to deploy a generative AI system that uses the Nuclear Regulatory Commission database and the Oak Ridge National Laboratory’s Frontier supercomputer to search data from Diablo Canyon’s files to synthesize information for federal regulatory filings. This use of AI is the first commercial use of generative AI at any nuclear power facility in the nation. The use of AI can have benefits; however, some recent demonstrations of AI have also shown that generative AI can also lead to unsafe and erroneous decisions. Several California gas and electric utilities are starting to integrate AI into efforts to prevent and respond to wildfires. Electric utilities have turned to AI solutions to obtain better data identifying infrastructure with greater risks for igniting wildfires and use of cameras with AI data analysis to identify potential fires before they grow. Early results from the use of these AI wildfire mitigation tools indicates that the tools are successful at reducing wildfire ignitions associated with utility infrastructure and provide early notifications to emergency personnel when fires are spotted.
Multiple agencies oversee utility safety requirements, but few standards exist for AI utility use. While AI can provide substantial resources to improve utility safety and reliability, AI can also pose risks to the energy sector. Biases in algorithms or poor AI training can lead to systemically reproduced biased and erroneous outcomes. Superintelligent AI systems are capable of creating systems that lack sufficient human input to stop harmful outcomes. The CPUC is one of several state and federal agencies that have adopted rules setting forth specific requirements for certain utility safety measures. Multiple CPUC general orders establish requirements for safe installation and operation of electric and natural gas infrastructure. Under existing utility safety rules, utilities under the CPUC’s jurisdiction must report certain critical incidents to the CPUC within two working hours or four non-working hours if the incident resulted in death, hospitalization, major news coverage, significant property damage, or prompted an investigation regarding potential utility wildfire ignition. Additionally, CPUC General Order 166 (GO 166) establishes standards for electric utilities’ operation during emergencies and disasters, which includes mandatory reporting regarding major power outages. While the CPUC has adopted specific requirements regarding safety measures for electrical and gas infrastructure, the CPUC has not adopted specific safety requirements for utilities’ use of AI when operating electrical and gas infrastructure. The CPUC generally has broad authority to regulate utilities to ensure the provision of safe and reliable gas and electric service at just and reasonable rates; however, it is unclear if CPUC rules have been sufficiently reviewed and updated to reflect the use of AI in utility operations. It is also not clear whether the CPUC has information from all the utilities regarding the types of AI models used. While some AI models may pose limited risks, other AI models that use more intelligent generative AI functions can pose greater risks by amplifying biases and creating data “hallucinations.” This bill would expand disclosures of electrical and gas utilities’ use of AI models and expand the CPUC’s authority to regulate these utilities’ use of AI.
Bill requires the CPUC to regulate human review of utilities’ AI operations. This bill requires the CPUC to establish standards for utilities’ AI models, and this bill requires those standards to include specific provisions regarding human review of AI processes in utility operations. Specifically, this bill requires the CPUC to establish requirements for human review during the deployment of AI models. AI models may differ widely from each other and may be developed for highly customized purposes. To effectively evaluate the appropriate level of human review for each use of AI models in the energy utility sector, the CPUC may need to evaluate AI deployment usage on a case-by-case basis to ensure that AI used for utility safety and reliability is not impeded while preventing harmful AI outcomes.
Related/Prior Legislation
SB 947 (McNerney) of 2026, limits employers’ use of ADS in certain employment decisions and requires the Labor Commissioner to oversee employers’ use of ADS in workplace decisions. The bill is pending in the Senate.
SB 833 (McNerney) of 2025, establishes requirements for human oversight over AI systems used with critical infrastructure. The bill also defined AI as including ADS. The bill is pending in the Assembly.
SB 420 (Padilla) of 2025, regulates the use of high-risk ADS, including requiring ADS developers and deployers to performing impact assessments on their systems. The bill establishes the right of individuals to know when an ADS has been used, details about the systems, and an opportunity to appeal ADS decisions, where technically feasible. The bill is pending in the Assembly.
SB 53 (Weiner, Chapter 138, Statutes of 2025) required large developers of certain AI tools to adopt protocols to address catastrophic risks associated with AI, disclose certain information about their AI models, and submit specified reports to Office of Emergency Services.
SB 7 (McNerney) of 2025, would have established limitations and notices for the use of ADS in employment decisions. The bill was vetoed.
AB 1018 (Bauer-Kahan) of 2025, would have regulated the use of ADS. It established obligations on developers and deployers of ADS designed or used to make or facilitate “consequential decisions. The bill is on the Senate Inactive File.
FISCAL EFFECT: Appropriation: No Fiscal Com.: Yes Local: Yes
According to the Senate Appropriations Committee, ongoing costs likely in the high hundreds of thousands of dollars annually (ratepayer funds) for the CPUC to adopt standards governing the use of AI models in utility operations, review and evaluate utility AI safety plans for compliance, establish oversight and enforcement mechanisms to ensure adherence to adopted standards, and assess and monitor any impacts to the provision of safe, affordable, and reliable electrical or gas service.
SUPPORT: (Verified 5/14/26)
California Federation Labor Unions, AFL-CIO (Co-source)
California School Employees Association, AFL-CIO (Co-source)
Engineers and Scientists of California, Local 20, IFPTE (Co-source)
Utility Workers Union of America, Local 132 (Co-source)
Utility Workers Union of America, Local 483 (Co-source)
Utility Workers Union of America, Local 522 (Co-source)
California Alliance for Retired Americans
Oakland Privacy
OPPOSITION: (Verified 5/14/26)
Burbank Water and Power
California Chamber of Commerce
California Municipal Utilities Association
Imperial Irrigation District
San Diego Gas and Electric Company
Southern California Gas Company
Southern California Public Power Authority
ARGUMENTS IN SUPPORT: According to the author:
AI has the potential to improve efficiency in California’s power grid and quicken response to wildfires and life-threatening dangers, but AI also presents significant risks. AI can hallucinate – make mistakes that could pose serious threats. Without human oversight, unchecked AI could trigger catastrophic equipment failures, service outages for thousands of residents, or life-threatening safety hazards. California can’t afford to outsource the safety of our power grid to unproven, unmonitored algorithms. California also needs to protect workers in the power sector from widespread job displacement. SB 1011 is a commonsense solution that will reduce risk by ensuring that AI use supports human decision-making rather than replacing it.
ARGUMENTS IN OPPOSITION: Opponents argue that the CPUC is not equipped to regulate AI models and regulations adopted by the CPUC may impede the use of AI models that are beneficial in the utility sector. In opposition, the California Chamber of Commerce states:
The bill’s definitions of “artificial intelligence model” is so broad that the bill likely sweeps in a wide range of longstanding utility systems, including distribution and outage management systems, powerplant control systems, and platforms used to manage distributed energy resources. In doing so, the bill risks inadvertently capturing decades of standardized utility technology already embedded in day-to-day grid and natural gas system operations.
The Southern California Public Power Authority (SCPPA) opposes this bill, arguing that requiring POUs to adopt policies similar to those imposed by the CPUC on IOUs would extend the CPUC’s regulatory powers to locally-governed POUs. In opposition, SCPPA states:
It should also be noted that SB 1011 requires all POUs to develop these CPUC policies, regardless of whether the utility is actually using AI models or already has an existing policy governing its use of AI models. In these cases, the bill is an unnecessary or duplicative paper exercise that adds an undue administrative burden on not-for-profit POUs that operate on a local government budget. The utilities represented by SCPPA are public agencies, operated by public employees, and governed by locally elected officials. Utility operational decisions are guided by what is in the best interest of the public they serve. This includes reliability, safety and cost.
Prepared by: Sarah Smith / E., U. & C. / (916) 651-4107
5/18/26 15:17:57
**** END ****
SENATE COMMITTEE ON
Senator Sabrina Cervantes, Chair
2025 - 2026 Regular Session
SB 1011 (McNerney) - Energy: Utility Infrastructure AI Safety, Oversight, and Workforce Protection Act
Version: | Policy Vote: |
Urgency: | Mandate: |
Hearing Date: | Consultant: |
Bill Summary:
This bill would require the California Public Utilities Commission (CPUC) to adopt standards for an electrical or gas corporation’s use of artificial intelligence (AI) models, as provided, and direct corporations to file a plan demonstrating compliance with those standards. This bill would also authorize the CPUC to prohibit use of an AI model if it finds that its deployment would negatively impact the provision of safe, affordable, and reliable electrical or gas service.
*********** ANALYSIS ADDENDUM – SUSPENSE FILE ***********
The following information is revised to reflect amendments
adopted by the committee on May 14, 2026
Fiscal Impact:
- Ongoing costs likely in the high hundreds of thousands of dollars annually (ratepayer funds) for the CPUC to adopt standards governing the use of AI models in utility operations, review and evaluate utility AI safety plans for compliance, establish oversight and enforcement mechanisms to ensure adherence to adopted standards, and assess and monitor any impacts to the provision of safe, affordable, and reliable electrical or gas service.
Committee Amendments:
These amendments would provide greater flexibility to the CPUC in regulating utility staffing related to AI. Specifically, committee amendments would require the CPUC to:
- Ensure that, for any job classification identified as responsible for human review and approval of an artificial intelligence model, electrical or gas corporations maintain staffing levels sufficient to support a meaningful human review and approval processes, as determined by the commission.
- Identify and establish requirements for human review and approval of any use of artificial intelligence that is high-risk, safety-sensitive, or would have a material operational impact.
- Remove some of the more proscriptive requirements, including that an electrical or gas corporation must demonstrate sufficient job classification personnel in a workforce plan and that AI models for certain applications must undergo human review and approval that meets specified criteria.
Providing additional flexibility to the CPUC can reduce state costs while still supporting and furthering the goals of utility infrastructure AI safety, oversight, and workforce protection. Committee amendments preserve importation protections while simplifying requirements placed on the CPUC and lowering costs for the development, implementation, and administration of standards for the use of AI models by electrical or gas corporations.
-- END --
SENATE COMMITTEE ON
Senator Sabrina Cervantes, Chair
2025 - 2026 Regular Session
SB 1011 (McNerney) - Energy: Utility Infrastructure AI Safety, Oversight, and Workforce Protection Act
Version: | Policy Vote: |
Urgency: | Mandate: |
Hearing Date: | Consultant: |
Bill Summary:
This bill would require the California Public Utilities Commission (CPUC) to adopt standards for an electrical or gas corporation’s use of artificial intelligence (AI) models, as provided, and to direct corporations to file a demonstrating it’s compliance with those standards. It would also authorize the CPUC to prohibit use of an AI model if it finds that deployment of the model would negatively impact the provision of safe, affordable, and reliable electrical or gas service.
Fiscal Impact:
- The CPUC estimates ongoing costs of about $1.4 million annually (ratepayer funds) to adopt standards governing the use of AI models in utility operations, review and evaluate utility AI safety plans for compliance, establish oversight and enforcement mechanisms to ensure adherence to adopted standards, and assess and monitor high-risk applications, including wildfire mitigation, grid design, and critical system operations.
Background:
Utilities’ use of generative AI is nascent; however, some utilities are seeking ways to develop AI tools that are trained to help with specific power sector needs. Diablo Canyon Power Plant partnered with Atomic Canyon to deploy a generative AI system that uses the Nuclear Regulatory Commission database and the Oak Ridge National Laboratory’s Frontier supercomputer to search data from Diablo Canyon’s files to synthesize information for federal regulatory filings. This use of AI is the first commercial use of generative AI at any nuclear power facility in the nation.
California’s electrical and natural gas infrastructure is essential to public safety, wildfire prevention, and economic stability. As investor-owned utilities and public utilities increasingly deploy AI systems, it is imperative to ensure there is human oversight and the professional judgment required to handle high-stakes situations, such as detecting wildfires, malfunctioning equipment, and natural gas leaks, and coordinating emergency response to disasters.
Proposed Law:
This bill would:
- Require the CPUC, as part of a new or existing proceeding, on or before January 1, 2028, to adopt standards for an electrical corporation’s or gas corporation’s use of artificial intelligence models. Require the standards adopted to do at least all of the following:
- Establish disclosure requirements for an electrical corporation or gas corporation to specify the types of artificial intelligence models it uses and how those models are used by the electrical corporation or gas corporation.
- Require an electrical corporation or gas corporation to identify the employee job classifications impacted by any planned implementation of an artificial intelligence model and efforts taken to ensure effective education, training, and retention for impacted employees.
- Identify steps an electrical corporation or gas corporation shall take to ensure that artificial intelligence models do not impact utility safety, affordability, and reliability.
- Establish requirements for human review and approval for the deployment and use of artificial intelligence models to ensure that electrical and gas services are safe, affordable, and reliable, including requirements for the placement of human review in the deployment of artificial intelligence models.
- Ensure that deployment of artificial intelligence models do not displace utility employees needed for the safe, affordable, and reliable provision of electrical and gas services. For any job classification identified as responsible for human review and approval of an artificial intelligence model, the electrical corporation or gas corporation shall demonstrate that its workforce plan maintains sufficient personnel in that job classification to ensure that human review and approval processes are not rendered nominal by understaffing.
- Establish requirements for human review and approval of any artificial intelligence model that does any of the following:
- Makes or directly implements operational decisions affecting the mapping, design, configuration, operation, maintenance, or oversight of electrical or gas infrastructure without a mandatory human approval step integrated into the operational workflow before implementation.
- Generates recommendations that the electrical corporation or gas corporation has configured to be automatically implemented without case-by-case human review.
- Performs any act that could directly and foreseeably result in physical harm to a person or a service interruption or outage, or impact on public safety.
- Provide requirements for human review and approval of artificial intelligence models to ensure staff conducting the review and approval meet specified criteria.
- Authorize the CPUC to prohibit an electrical corporation’s or gas corporation’s use of an artificial intelligence model if the commission finds that deployment of the artificial intelligence model would negatively impact the provision of safe, affordable, and reliable electrical or gas service.
- Require the CPUC to direct an electrical corporation or gas corporation to file a plan that demonstrates the corporation’s compliance with the standards adopted pursuant to this section. Allow the CPUC to request records as it deems necessary to verify compliance.
- Require the CPUC to ensure that a plan filed by an electrical corporation may be reviewed by the Office of Energy Infrastructure Safety to ensure that any artificial intelligence model used for wildfire mitigation is consistent with the electrical corporation’s wildfire mitigation plan.
- Require the CPUC to ensure that impacted bargaining units are consulted in the development of plans filed.
- Require each community choice aggregator and local publicly owned electric utility to adopt a policy regarding its use of an AI model consistent with the standards adopted by the CPUC, as provided.
Related Legislation:
SB 947 (McNerney) of 2026, limits employers’ use of ADS in certain employment decisions and requires the Labor Commissioner to oversee employers’ use of ADS in workplace decisions.
SB 833 (McNerney) of 2025, establishes requirements for human oversight over AI systems used with critical infrastructure. The bill also defined AI as including ADS. The bill is pending in the Assembly.
SB 420 (Padilla) of 2025, regulates the use of high-risk ADS, including requiring ADS developers and deployers to performing impact assessments on their systems. The bill establishes the right of individuals to know when an ADS has been used, details about the systems, and an opportunity to appeal ADS decisions, where technically feasible. The bill is pending in the Assembly.
SB 53 (Weiner, Chapter 138, Statutes of 2025) required large developers of certain AI tools to adopt protocols to address catastrophic risks associated with AI, disclose certain information about their AI models, and submit specified reports to Office of Emergency Services.
SB 7 (McNerney) of 2025, would have established limitations and notices for the use of ADS in employment decisions. The bill was vetoed.
AB 1018 (Bauer-Kahan) of 2025, would have regulated the use of ADS. It established obligations on developers and deployers of ADS designed or used to make or facilitate “consequential decisions. The bill is on the Senate Inactive File.
-- END --
SENATE PRIVACY, DIGITAL TECHNOLOGIES, AND CONSUMER PROTECTION COMMITTEE
Senator Christopher Cabaldon, Chair
2025-2026 Regular Session
Bill No: SB 1011 (Author:McNerney)
Version: February 10, 2026
Hearing Date: April 20, 2026
Fiscal: Yes
Urgency: No
Consultant: CK
SUBJECT
Energy: Utility Infrastructure AI Safety, Oversight, and Workforce Protection Act
DIGEST
This bill requires the California Public Utilities Commission (CPUC) to adopt standards for electrical and gas corporations’ use of AI models, as specified.[1]
EXECUTIVE SUMMARY
California’s electrical and natural gas infrastructure is essential to public safety, wildfire prevention, and economic stability. As investor-owned utilities and public utilities increasingly deploy AI systems, it is imperative to ensure there is human oversight and the professional judgment required to handle high-stakes situations, such as detecting wildfires, malfunctioning equipment, and natural gas leaks, and coordinating emergency response to disasters.
This bill tasks the CPUC with adopting standards for electrical and gas corporations’ use of AI systems that include specified elements. The CPUC must require these corporations to file plans demonstrating compliance and is authorized to prohibit use of an AI model if it finds it would negatively impact service provision. Community choice aggregators and local publicly owned electric utilities are also required to adopt policies for utility use of AI that are consistent with the CPUC standards.
This bill is sponsored by the California Federation of Labor Unions and the Utility Workers Union of America Locals 132, 483, and 522. The bill is supported by Oakland Privacy and the California School Employees Association. It is opposed by local government associations and the California Chamber of Commerce. This bill passed out of the Senate Energy, Utilities, and Communications Committee on a vote of 13 to 3.
PROPOSED CHANGES TO THE LAW
Existing law:
- Requires the CPUC to establish rules for a public utility and revise those requirements through an order or rule whenever the CPUC, after a hearing, finds that existing rules, practices, equipment, appliances, facilities or service of any public utility or manufacturing, distribution, transmission, storage or supply methods employed by the public utility are unjust, unreasonable, unsafe, improper, inadequate, or insufficient. (Pub. Util. Code § 761.)
- Requires every public utility to establish and maintain adequate, efficient, just, and reasonable services, instruments, equipment and facilities needed to promote the health, safety and comfort of customers, employees, and the public. Any rule adopted by a public utility regarding its services or rates for the public must be just and reasonable. (Pub. Util. Code § 451)
- Establishes the Office of Energy Infrastructure Safety in the Natural Resources Agency to oversee electrical corporations’ compliance with wildfire safety requirements, as specified. (Gov. Code § 15470 et. seq.)
- Requires the California Department of Technology (CDT) to conduct a comprehensive inventory of all high-risk automated decision systems (ADS) that have been proposed for use, development, or procurement by, or are being used, developed, or procured by, any state agency. It defines the relevant terms:
- “Automated decision system” means a computational process derived from machine learning, statistical modeling, data analytics, or AI that issues simplified output, including a score, classification, or recommendation, that is used to assist or replace human discretionary decisionmaking and materially impacts natural persons. “Automated decision system” does not include a spam email filter, firewall, antivirus software, identity and access management tools, calculator, database, dataset, or other compilation of data.
- “High-risk automated decision system” means an ADS that is used to assist or replace human discretionary decisions that have a legal or similarly significant effect, including decisions that materially impact access to, or approval for, housing or accommodations, education, employment, credit, health care, and criminal justice. (Gov. Code § 11546.45.5.)
- Defines “AI” as an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments. (Gov. Code § 11546.45.5.)
This bill:
- Establishes the Utility Infrastructure AI Safety, Oversight, and Workforce Protection Act.
- Requires the CPUC, as part of a new or existing proceeding, on or before January 1, 2028, to adopt standards for electrical and gas corporations’ use of AI models.
- Requires these standards to achieve at least all of the following:
- Establish disclosure requirements for an electrical or gas corporation to specify the types of AI models it uses and how these models are used by the electrical and gas corporation.
- Require an electrical or gas corporation to identify the employee job classifications impacted by any planned implementation of an AI model and efforts taken to ensure effective education and training and retention for impacted employees.
- Identify steps an electrical or gas corporation shall take to ensure that AI models do not impact utility safety, affordability, and reliability.
- Establish requirements for human review and approval for the deployment and use of AI models to ensure that electrical and gas services are safe, affordable, and reliable, including requirements for the placement of human review in the deployment of AI models. Standards for human review and approval of AI models shall ensure staff conducting the review and approval meet all the following:
- Have relevant expertise in the operational area affected by the AI model.
- Have been provided sufficient information, time, and access to data to meaningfully evaluate the model’s output or recommendations.
- Retain unimpeded authority to reject, modify, or defer implementation without adverse consequence to the reviewer.
- Have not been subject to a workflow design, performance metric, or operational pressure that renders rejection of the system output functionally impractical.
- Ensure that deployment of AI models does not displace utility employees needed for the safe, affordable, and reliable provision of electrical and gas services. For any job classification identified as responsible for human review and approval for an AI model, the electrical or gas corporation shall demonstrate that its workforce plan maintains sufficient personnel in that classification to ensure that human review and approval processes are not rendered nominal by understaffing.
- Establish requirements for human review and approval of any AI system that does any of the following:
- Makes or directly implements operational decisions affecting the mapping, design, configuration, operation, maintenance, or oversight of electrical or gas infrastructure without a mandatory human approval step integrated into the operational workflow prior to implementation.
- Generates recommendations that the covered electrical or gas corporation has configured to be automatically implemented without case-by-case human review.
- Performs any act that could directly and foreseeably result in physical harm to persons, a service interruption or outage, or an impact on public safety.
- Authorizes the CPUC to prohibit an electrical or gas corporation’s use of an AI model if the commission finds that deployment of the model would negatively impact the provision of safe, affordable, and reliable electrical or gas service.
- Requires the CPUC to direct an electrical or gas corporation to file plans that demonstrate the corporation’s compliance with the adopted standards. The CPUC may request records as it deems necessary to verify compliance. The CPUC must ensure that impacted bargaining units are consulted in the development of these plans.
- Requires the CPUC to ensure that plans filed by an electrical corporation may be reviewed by the Office of Energy Infrastructure Safety to ensure that any AI model used for wildfire mitigation is consistent with the electrical or gas corporation’s wildfire mitigation plan, as provided.
- Requires community choice aggregators and local publicly-owned electric utilities to adopt policies regarding the utility’s use of AI models that are consistent with the CPUC standards.
- Clarifies that it establishes minimum safety and governance standards of AI models and does not limit, waive, or alter any rights, remedies, or obligations under state or federal law or any collective bargaining agreement, with respect to technological change, staffing, workload, training, or working conditions.
- Includes a severability clause.
COMMENTS
- The risks presented by AI models and applications
With recent dramatic advances in the capabilities of AI systems, the need for frameworks for accountability and responsible development has become ever more urgent.
As directed by the National AI Initiative Act of 2020, the National Institute of Standards and Technology (NIST) developed the AI Risk Management Framework to assist entities designing, developing, deploying, and using AI systems to help manage the many risks of AI and promote trustworthy and responsible development and use of AI systems. That framework highlights the serious risks at play and the uniquely challenging nature of addressing them in this context:
Artificial intelligence (AI) technologies have significant potential to transform society and people’s lives – from commerce and health to transportation and cybersecurity to the environment and our planet. AI technologies can drive inclusive economic growth and support scientific advancements that improve the conditions of our world. AI technologies, however, also pose risks that can negatively impact individuals, groups, organizations, communities, society, the environment, and the planet. Like risks for other types of technology, AI risks can emerge in a variety of ways and can be characterized as long- or short-term, high or low-probability, systemic or localized, and high- or low-impact.
While there are myriad standards and best practices to help organizations mitigate the risks of traditional software or information-based systems, the risks posed by AI systems are in many ways unique. AI systems, for example, may be trained on data that can change over time, sometimes significantly and unexpectedly, affecting system functionality and trustworthiness in ways that are hard to understand. AI systems and the contexts in which they are deployed are frequently complex, making it difficult to detect and respond to failures when they occur. AI systems are inherently socio-technical in nature, meaning they are influenced by societal dynamics and human behavior. AI risks – and benefits – can emerge from the interplay of technical aspects combined with societal factors related to how a system is used, its interactions with other AI systems, who operates it, and the social context in which it is deployed.
These risks make AI a uniquely challenging technology to deploy and utilize both for organizations and within society. [. . .]
AI risk management is a key component of responsible development and use of AI systems. Responsible AI practices can help align the decisions about AI system design, development, and uses with intended aim and values. Core concepts in responsible AI emphasize human centricity, social responsibility, and sustainability. AI risk management can drive responsible uses and practices by prompting organizations and their internal teams who design, develop, and deploy AI to think more critically about context and potential or unexpected negative and positive impacts. Understanding and managing the risks of AI systems will help to enhance trustworthiness, and in turn, cultivate public trust.
This highlights how the risks posed by AI are inherently complex and ever-changing. Constant adaptations and nimble responses to addressing potential risks are of critical importance.
More recently, the Biden Administration published its Blueprint for an AI Bill of Rights, which is a set of five principles and associated practices to help guide the design, use, and deployment of AI to protect the rights of the American public. One key piece focuses on the safety of these systems: “Safe and Effective Systems: You should be protected from unsafe or ineffective systems. Automated systems should be developed with consultation from diverse communities, stakeholders, and domain experts to identify concerns, risks, and potential impacts of the system.”[2]
While the future is unclear, the need to respond to these potential harms now is evident. The Center for New American Security puts a fine point on it:
While there is significant uncertainty in how the future of AI develops, current trends point to a future of vastly more powerful AI systems than today’s state of the art. The most advanced systems at AI’s frontier will be limited initially to a small number of actors but may rapidly proliferate. Policymakers should begin to put in place today a regulatory framework to prepare for this future. Building an anticipatory regulatory framework is essential because of the disconnect in speeds between AI progress and the policymaking process, the difficulty in predicting the capabilities of new AI systems for specific tasks, and the speed with which AI models proliferate today, absent regulation. Waiting to regulate frontier AI systems until concrete harms materialize will almost certainly result in regulation being too late.[3]
- Establishing a framework for AI deployment in the energy sector
According to the author:
AI has the potential to improve efficiency in California’s power grid and quicken response to wildfires and life-threatening dangers, but AI also presents significant risks. AI can hallucinate – make mistakes that could pose serious threats. Without human oversight, unchecked AI could trigger catastrophic equipment failures, service outages for thousands of residents, or life-threatening safety hazards. California can’t afford to outsource the safety of our power grid to unproven, unmonitored algorithms. California also needs to protect workers in the power sector from widespread job displacement. SB 1011 is a commonsense solution that will reduce risk by ensuring that AI use supports human decision-making rather than replacing it.
This bill obligates the CPUC, as part of a new or existing proceeding, on or before January 1, 2028, to adopt standards for electrical and gas corporations’ use of AI models. The bill lays out a series of elements that must be included within these standards. This includes establishing clear disclosure requirements about the type of AI models used and how they are used, and requirements to identify the job classifications impacted and the efforts taken to ensure effective education, training, and retention. The standards must also identify required steps to ensure AI does not impact utility safety, affordability, and reliability. There must be requirements around human review and approval. The standards must ensure that AI deployment does not displace employees necessary for the proper provision of electrical or gas services.
The bill also authorizes the CPUC to prohibit AI use if it finds that such deployment would negatively impact the proper provision of the relevant services. The CPUC is obligated to direct electrical and gas corporations to file plans that demonstrate compliance with the above standards and authorized to request records necessary to verify such compliance. Impacted bargaining units must be consulted in development of these plans.
Specific to wildfire safety, the CPUC must ensure that plans filed by an electrical corporation may be reviewed by the Office of Energy Infrastructure Safety to ensure that any AI model used for wildfire mitigation is consistent with the electrical or gas corporation’s wildfire mitigation plan.
Community choice aggregators and local, publicly-owned electrical utilities are required to adopt policies regarding their use of an AI model that is consistent with the CPUC standards.
The California Federation of Labor Unions, a co-sponsor of the bill, explains the need for this legislation:
Utilities are currently using AI tools with the goal of reducing hours-long document search tasks and helping staff focus on higher-value analysis. While this kind of AI application is described as necessary for improvements to efficiency, it raises concerns about error risks in critical environments, and future expansion into core operational decision-making without appropriate governance. Such risks include incorrect or misleading outputs, often referred to as “hallucinations,” particularly when AI systems operate outside their training data or encounter novel conditions. In utility operations, even small errors in mapping, system records, or control recommendations can cascade into serious reliability or safety events such as the failure to detect or prevent devastating wildfires or natural gas leaks.
The rapid deployment of automated systems raises concerns about the displacement of experienced, licensed engineers, technical specialists, and field personnel whose expertise is essential to safe utility operations. Without clear statewide standards for human oversight, testing, transparency, and workforce protections, unchecked AI deployment could undermine system reliability, weaken safety controls, and erode the skilled workforce responsible for operating California’s critical infrastructure.
Registered opposition focuses on definitions and provisions governing the use of ADS. The bill, as agreed to be amended and as analyzed herein, is an entirely new approach. Many of those in opposition have indicated that they are still processing these recent amendments.
SUPPORT
California Labor Federation, AFL-CIO (co-sponsor)
Utility Workers Union of America Locals 132, 483, and 522 (co-sponsor)
California School Employees Association
Oakland Privacy
OPPOSITION
Burbank Water and Power
California Chamber of Commerce
California Municipal Utilities Association
Southern California Public Power Authority
RELATED LEGISLATION
SB 833 (McNerney, 2026) requires an oversight personnel for an operator that deploys a covered AI system to establish a human oversight mechanism that ensures a human monitors the AI system’s operations in real time and reviews and approves any plan or action proposed by an AI system before execution. “Operator” means a state agency responsible for operating, managing, overseeing, or controlling access to critical infrastructure. SB 833 is currently in the Assembly Appropriations Committee.
SB 947 (McNerney, 2026) regulates the use of ADS in the employment context by requiring employers to provide postuse notices that inform workers that they are subject to ADS and of the ADS details. It provides a series of prohibited uses, such as where it may interfere with existing labor protections or where it conducts predictive behavior analysis, as defined. Workers have the right to access information used by the ADS, including inputs, outputs, and any corroborating evidence used by the employer in making the decision. SB 947 is set to be heard the same day as this bill in this Committee.
PRIOR VOTES:
Senate Energy, Utilities and Communications Committee (Ayes 13, Noes 3)
**************
Extensive amendments were agreed to in the previous committee with the amendments to be taken in this Committee. This analysis is of the bill as agreed to be amended. ↑
Blueprint For An AI Bill Of Rights (October 2022) Office of Science and Technology Policy, https://www.whitehouse.gov/wp-content/uploads/2022/10/Blueprint-for-an-AI-Bill-of-Rights.pdf. All internet citations are current as of April 16, 2026. ↑
Paul Scharre, Future-Proofing Frontier AI Regulation (March 2024) Center for New American Security, https://s3.us-east-1.amazonaws.com/files.cnas.org/documents/CNAS-Report_AI-Trends_FinalC.pdf. This, and all further, internet citations are current as of June 14, 2025. ↑
SENATE COMMITTEE ON ENERGY, UTILITIES AND COMMUNICATIONS
Senator Benjamin Allen, Chair
2025 - 2026 Regular
Bill No: Hearing Date: 4/13/2026
Author: | |||
Version: | Introduced | ||
Urgency: | Fiscal: | ||
Consultant: | |||
SUBJECT:
DIGEST: This bill limits electrical and gas utilities’ ability to use automated decision systems (ADS) in certain aspects of utility operations, as specified. This bill requires utilities to enable qualified utility personnel to override or modify outcomes from an ADS, and sets standards for requiring human review and approval for use of an ADS in certain circumstances, including, but not limited to requiring a licensed engineer to oversee the use of an automated decision system in any circumstance that meets this bill’s definition of an ”engineering decision.” This bill requires a utility to provide notices to labor organizations and affected employees before implementing an ADS that impacts workers’ duties, training, or staffing levels. This bill requires the California Public Utilities Commission (CPUC) to regulate an investor-owned utility (IOU) use of an ADS and requires the California Energy Commission (CEC) to regulate a publicly owned utility (POU) use of an ADS.
ANALYSIS:
Existing law:
- Establishes the CPUC, consisting of five members appointed by the Governor, and authorizes the CPUC to fix rates and establish rules for public utilities. (Article XII of the California Constitution)
- Requires the CPUC to establish rules for a public utility and revise those requirements through an order or rule whenever the CPUC, after a hearing, finds that existing rules, practices, equipment, appliances, facilities or service of any public utility or manufacturing, distribution, transmission, storage or supply methods employed by the public utility are unjust, unreasonable, unsafe, improper, inadequate, or insufficient. (Public Utilities Code §761)
- Requires every public utility to establish and maintain adequate, efficient, just, and reasonable services, instruments, equipment and facilities needed to promote the health, safety and comfort of customers, employees, and the public. Any rule adopted by a public utility regarding its services or rates for the public must be just and reasonable. (Public Utilities Code §451)
- Establishes the California Wildfire Safety Advisory Board, consisting of seven members to advise utilities and the Office of Energy Infrastructure Safety on utility wildfire safety issues, as specified. (Public Utilities Code §326.1)
- Establishes the Office of Energy infrastructure Safety in the Natural Resources Agency to oversee electrical corporations’ compliance with wildfire safety requirements, as specified. (Government Code §15470 et. seq.)
- Establishes the CEC, consisting of five members appointed by the Governor, and specifies the duties of the CEC. Every two years, the Governor must designate a chair and vice chair from the CEC’s membership. The CEC must appoint a public adviser every three years to carry out certain public engagement duties. (Public Resources Code §25200 et. seq.)
- Assigns various duties to the CEC, including but not limited to, duties regarding energy research and development, clean transportation infrastructure, powerplant siting, energy demand and supply analysis, energy efficiency standards, and administration of the renewable portfolio standard. (Public Resources Code §25000 et. seq.)
- Requires the CEC to assess trends in energy consumption and analyze the social, economic, and environmental consequences of these trends. The CEC must establish energy conservation measures, including building and appliance energy efficiency standards, and recommend additional conservation measures to the Governor and the Legislature. (Public Resources Code §25216)
- Defines an “artificial intelligence model” as an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments. Existing law establishes requirements for large developers of artificial intelligence models. (Business and Professions Code §22757.10 et. seq.)
This bill:
- Defines an “automated decision system” as a computational process, including one derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, and that is used to assist or replace human discretionary decision-making and materially impacts natural persons.
- Specifies that the bill’s definition of an “automated decision system” does not include the following:
- Spam email filters.
- Firewalls.
- Antivirus software.
- Identity and access management tools.
- Calculators.
- Databases, datasets, or other compilations of data.
- Defines an “engineering decision” as a decision, recommendation, or configuration change affecting the design, operation, maintenance, or configuration of electrical circuits, substations, gas pipelines, compressor stations, or other utility infrastructure that is subject to review by, or is required to be performed by, a person licensed under the Professional Engineers Act.
- Defines a “high-risk automated decision system” as an ADS used by a utility that does any of the following:
- Creates or modifies records, including geographic information system (GIS) layers, asset registers, and configuration logs, used in operational or engineering decision-making.
- Issues, populates, or prioritizes operational actions, such as switching, fault isolation, circuit reconfiguration, load transfers, or gas-flow adjustments.
- Prioritizes specified safety critical or wildfire risk mitigation decisions, line de-energization, vegetation clearance, and gas leak or electrical hazard classification.
- Makes recommendations affecting whether or when emergency response or field personnel dispatch is initiated.
- Generates, prioritizes, or recommends supervisory control and data acquisition.
- Control room or operational control actions that alter operating states, including set points, valve lineups, compressor controls, pressure or flow adjustments, alarms, interlocks, or operating limits.
- Creates, modifies, or recommends changes to certain safety critical data that governs utility operations.
- Designates the CPUC as the regulator of ADS use by an IOU and requires the CEC to regulate the use of an ADS by a POU.
- Requires every electrical or gas utility using an ADS to maintain a process by which qualified personnel can override or change the output of the ADS.
- Prohibits electrical and gas utilities from using a high-risk ADS without a human reviewing and approving the outcome produced by the ADS.
- Requires electrical and gas utilities to obtain human review and approval from a licensed engineer in circumstances where a high-risk ADS would produce an outcome that constitutes an engineering decision. This bill prohibits electrical and gas utilities from using high-risk ADS systems that can independently execute actions or update records.
- Requires electrical and gas utilities to maintain a log of all ADS actions with specified data.
- Prohibits electrical and gas utilities from using high risk ADS systems without first filing a specified safety plan with the CPUC or CEC. An electrical or gas utility must provide a copy of this safety plan to affected labor organizations at the time it files the plan with the CPUC or CEC – or within 15 days of receiving a request.
- Requires an electrical or gas utility to test the use of an ADS in a staging mode for 18 months, or a time determined by the CPUC or CEC. This staging mode must include human auditing and approval for all changes before the ADS can be used in utility operations.
- Establishes specified requirements for a high-risk ADS used for certain modeling and recordkeeping, including maintaining an audit trail of all changes from the ADS and human review for an unspecified number of years. This bill requires the utility to make this audit trail available to the CEC or CPUC upon request.
- Requires electrical and gas utilities to provide at least 180 days notice to affected labor organizations and impacted employee classifications before introducing any ADS that materially impacts utility workers’ job duties, classifications, staffing levels, or required training. This bill specifies items that must be included in this notice
- Prohibits electrical and gas utilities from deploying a high-risk ADS that results in layoffs for utility engineering workers, mapping, design or technical operations until the utility exhausts retaining, redeployment, or reclassification options.
Background
ADS encompass many activities – not just AI. The use of AI can have benefits; however, some recent demonstrations of AI have also shown that generative AI can also lead to unsafe and erroneous decisions. Biases in algorithms or poor AI training can lead to systemically reproduced biased and erroneous outcomes. Superintelligent AI systems are capable of creating systems that lack sufficient human input to stop harmful outcomes. While this bill is aimed at preventing the proliferation of AI decisions that may impact utility safety and jobs of utility employees, this bill does not solely limit actions that would fall under the umbrella artificial intelligence models. This bill establishes limitations on electrical and gas utilities’ use of ADS and establishes requirements for the use of ADS in certain circumstances. ADS encompasses a broad number of systems, including automated actions and technologies that do not use any artificial intelligence models. This bill’s definition of ADS may encompass the use of AI models; however, this definition may also include many automated systems and technologies that have long been used by utilities to help utilities respond to grid conditions, market signals, and safety needs in real time. By incorporating a broad definition of ADS, this bill may impact a wide variety of utility operations.
How do utilities use AI and ADS? Utilities’ use of generative AI is nascent; however, some utilities are seeking ways to develop AI tools that are trained to help with specific power sector needs. Diablo Canyon Power Plant partnered with Atomic Canyon to deploy a generative AI system that uses the Nuclear Regulatory Commission database and the Oak Ridge National Laboratory’s Frontier supercomputer to search data from Diablo Canyon’s files to synthesize information for federal regulatory filings. This use of AI is the first commercial use of generative AI at any nuclear power facility in the nation.
While some utilities are exploring options for using generative AI, the vast majority of automated systems used by utilities are not generative AI. While some ADS technologies are implemented to improve safety and enhance reliability, others are necessary to participate in real-time energy markets. For example, automatic reclosers on power lines can help detect faults on electrical lines and shut off power to ensure the fault does not continue. When the recloser senses that the fault is temporary and not-recurring, the recloser re-sets and allows power to return. Similarly, gas utilities can use automatic shut-off valves (ASVs), which will automatically respond to pressure changes in a pipeline or work with seismic sensors to shut off the flow of gas in the event of an earthquake. Utilities are also increasingly using automatic sensors and predictive modeling tools as part of wildfire threat modeling. Smoke mapping can help utilities project the path of a fire’s spread and anticipate the need for additional mitigation and evacuation strategies. Electric utilities use Supervisory Control and Data Acquisition (SCADA) systems that use automatic sensors and data synthesis to help monitor real time conditions on the grid, including voltage levels. These SCADA systems are also used by utilities to participate in the Energy Imbalance Market because they provide the information, speed and resources necessary to transact energy in real-time to ensure a balance between supply and demand of resources.
As currently written, this bill’s definition of a “high-risk” ADS would apply to a utility’s use of reclosers, ASVs, wildfire modeling, and SCADA systems. This bill does not clarify whether its restrictions on ADS apply to already-operating systems. By requiring the utility to operate these systems in a staging mode for 18 months before using these existing systems, this bill could limit or delay the use of essential utility tools needed to safely and reliably operate. This bill also requires human approval for every output of a high-risk ADS, which may prevent utilities from using automatic reclosers, ASVs, or other technologies aimed at responding to real-time events to reduce injuries and ignitions from utility infrastructure. Limiting the use of these resources may prevent utilities from complying with existing wildfire mitigation plans and utility safety standards.
Multiple agencies oversee utility safety requirements, but few standards exist for AI utility use. The CPUC is one of several state and federal agencies that have adopted rules setting forth specific requirements for certain utility safety measures. Multiple CPUC general orders establish requirements for safe installation and operation of electric and natural gas infrastructure. Under existing utility safety rules, utilities under the CPUC’s jurisdiction must report certain critical incidents to the CPUC within two working hours or four non-working hours if the incident resulted in death, hospitalization, major news coverage, significant property damage, or prompted an investigation regarding potential utility wildfire ignition. Additionally, CPUC General Order 166 (GO 166) establishes standards for electric utilities’ operation during emergencies and disasters, which includes mandatory reporting regarding major power outages. This bill would establish a 24-hour reporting requirement if an ADS-related outcome resulted in the following:
- An outage or service interruption affecting more than 500 customers.
- Equipment damage, a failure, or a safety hazard.
- A mis-switching event, incorrect mapping or asset register update, erroneous configuration change, or data-integrity breach with safety or reliability implications.
This bill’s notification requirements may be more permissive than current emergency response requirements and major incident notification rules under existing CPUC general orders. While this bill attempts to establish standards for preventing and responding to incidents stemming from AI or ADS use, it is not clear that the standards would provide greater protections for utility safety than existing rules. However, it is possible that the CPUC may need to modify existing safety standards and disclosures if utilities intend to integrate AI into more utility operations.
Need for Amendments. As currently written, this bill’s definition of an automated decision system potentially encompasses and limits a utility’s ability to use systems intended to ensure stability of grid operations, reliable electricity supply, effective participation in certain energy market functions, and effective implementation of utility safety requirements. Additionally, this bill assigns regulatory oversight of utilities’ use of automated decision systems to the CEC, which does not currently have any oversight duties related to POU and IOU utility safety, workforce needs, or rates. Portions of this bill may also conflict with existing rules established by the CPUC for preventing, mitigating, and reporting utility safety incidents. For these reasons, the author and committee may wish to amend this bill to do the following:
- Remove requirements for the CEC to administer this bill.
- Require the CPUC to adopt standards for utilities’ use of artificial intelligence models.
- Conform this bill’s definition of an artificial intelligence model to definitions in existing law.
- Specify that the CPUC’s standards must include at least the following:
- Requirements for electrical corporations to specify the types of artificial intelligence models used and how those models are used.
- Identification of employee job classifications impacted by the planned implementation of any artificial intelligence models and efforts taken to ensure effective education and training and retention for impacted workers.
- Measures the utility has adopted to ensure that the models used will not impact utility safety, affordability, and reliability requirements.
- Requirements for human approval for AI deployment to ensure the safe, affordable and reliable operation of electrical and gas services.
- Ensure that AI deployment is not intended to displace utility workers needed for the safe, affordable and reliable provision of services.
- Require the CPUC to direct electrical corporations to file plans disclosing their use of artificial intelligence models and compliance with CPUC standards for these models.
- Specify that the CPUC may prohibit the use of an artificial intelligence model if the CPUC finds that deployment of the model would negatively impact the provision of safe, reliable, and affordable electrical or gas service.
- Require the CPUC to ensure that plans filed by the IOUs are provided to the Office of Energy Infrastructure Safety to ensure consistency with IOUs’ wildfire mitigation plans.
- Require boards of POUs to adopt plans consistent with the standards established by the CPUC.
Dual Referral. Should this bill be approved by this committee, it will be re-referred to the Senate Privacy, Digital Technologies, and Consumer Protection Committee.
Prior/Related Legislation
SB 947 (McNerney) of 2026, limits employers’ use of ADS in certain employment decisions and requires the Labor Commissioner to oversee employers’ use of ADS in workplace decisions. The bill is pending in the Senate Privacy, Digital Technologies, and Consumer Protection Committee.
SB 833 (McNerney) of 2025, establishes requirements for human oversight over AI systems used with critical infrastructure. The bill also defined AI as including ADS. The bill is pending in the Assembly.
SB 420 (Padilla) of 2025, regulates the use of high-risk ADS, including requiring ADS developers and deployers to performing impact assessments on their systems. The bill establishes the right of individuals to know when an ADS has been used, details about the systems, and an opportunity to appeal ADS decisions, where technically feasible. The bill is pending in the Assembly.
SB 53 (Weiner, Chapter 138, Statutes of 2025) required large developers of certain AI tools to adopt protocols to address catastrophic risks associated with AI, disclose certain information about their AI models, and submit specified reports to Office of Emergency Services.
SB 7 (McNerney) of 2025, would have established limitations and notices for the use of ADS in employment decisions. The bill was vetoed.
AB 1018 (Bauer-Kahan) of 2025, would have regulated the use of ADS. It established obligations on developers and deployers of ADS designed or used to make or facilitate “consequential decisions. The bill is on the Senate Inactive File.
FISCAL EFFECT: Appropriation: Fiscal Com.: Local:
SUPPORT:
California Federation Labor Unions, AFL-CIO (Co-Sponsor)
California School Employees Association, AFL-CIO (Co-Sponsor)
Engineers and Scientists of California, Local 20, IFPTE (Co-Sponsor)
Utility Workers Union of America, Local 132 (Co-Sponsor)
Utility Workers Union of America, Local 483 (Co-Sponsor)
Utility Workers Union of America, Local 522 (Co-Sponsor)
OPPOSITION:
Burbank Water and Power
California Municipal Utilities Association
San Diego Gas and Electric Company, unless amended
Southern California Gas Company, unless amended
Southern California Public Power Authority
ARGUMENTS IN SUPPORT: According to the author:
AI has the potential to improve efficiency in California’s power grid and quicken response to wildfires and life-threatening dangers, but AI also presents significant risks. AI can hallucinate – make mistakes that could pose serious threats. Without human oversight, unchecked AI could trigger catastrophic equipment failures, service outages for thousands of residents, or life-threatening safety hazards. California can’t afford to outsource the safety of our power grid to unproven, unmonitored algorithms. California also needs to protect workers in the power sector from widespread job displacement. SB 1011 is a commonsense solution that will reduce risk by ensuring that AI use supports human decision-making rather than replacing it.
ARGUMENTS IN OPPOSITION: Opponents argue that this bill would limit the use of a wide variety of measures and utility systems that provide reliability, safety and wildfire mitigation services. In opposition, the California Municipal Utilities Association states:
The extensive regulatory requirements contained in SB 1011 would function as a de-facto ban on the use of Automated Decision Systems (ADS) by POUs. Such a ban will negatively impact a POU’s ability to manage the electrical grid, mitigate wildfire risk, and improve overall efficiency, reliability and reduce ratepayer costs…Furthermore, we are concerned about the impacts the bill would have on utility grid operations and wildfire mitigation activities – all of which use some form of ADS currently and have for many years. For example, while the bill lists exclusions to these ADS compliance rules for things such as spam filters, firewalls, and basic IT security tools, it does not provide exclusions for grid and operational management platforms commonly used in utility operations and does not take into account wildfire and outage mitigation tools, such as automatic reclosers. We are concerned that utility operations would be hindered by this bill because it does not account for the nuances of our existing automated programs and systems.
-- END --
Discussed in Hearing
Senate Standing Committee on Privacy, Digital Technologies, and Consumer Protection
Senate Standing Committee on Energy, Utilities and Communications
Bill Author
Bill Co-Author(s):