EPRI has launched a new industry initiative aimed at helping electric utilities and technology providers evaluate the operational risks of artificial intelligence before deploying AI in critical grid applications.

The initiative, called SAFERai.power, builds on work from EPRI's Open Power AI (OPAI) consortium and is designed to establish a common framework for assessing AI systems used in utility operations. The effort focuses on ensuring AI deployments support key operational priorities, including safety, accountability, fairness, explainability and reliability.

According to EPRI, the initiative will develop an open-source risk assessment toolkit and standardized evaluation methods to help utilities determine whether AI applications are ready for use in operational environments.

"AI is ready to transform how the power system operates—but only if we make it trustworthy in the real world," said Arshad Mansoor, EPRI's president and CEO. "With SAFERai.power, we are building the bridge between innovation and trust, applying guiding principles to the tools needed for operations."

The initiative will produce three primary resources:

• A risk framework outlining AI risk scenarios, failure modes, autonomy levels, guardrails and use-case tiers.
• Evidence packages containing templates for documenting AI system risks from both technology providers and utilities deploying the systems.
• An open-source assessment toolkit that supports scalable, tiered risk evaluations for different AI use cases.

More than 25 organizations are participating as founding members, including utilities, technology companies, regulators, independent system operators and reliability organizations.

Several industry leaders said the collaborative approach is intended to help establish consistent practices as AI adoption accelerates across the electric sector.

Darryl Willis, corporate vice president of Energy and Resources at Microsoft Frontier Company, said trust will be essential as utilities deploy AI in critical infrastructure. He noted that AI systems must be transparent, secure and grounded in industry expertise to support reliable grid operations.

Regulators also see value in developing common guidance for AI deployment. Ann Rendahl, president of the National Association of Regulatory Utility Commissioners (NARUC), said collaborative efforts such as SAFERai.power can help establish best practices that support innovation while maintaining public confidence in the electric system.

Reliability organizations are also contributing to the initiative. Jim Robb, president and CEO of the North American Electric Reliability Corporation (NERC), said understanding both the benefits and risks of AI will be critical as the technology becomes more widely adopted across utility operations.

Robert Piascik, senior vice president and chief information officer at the New York Power Authority, said utilities need evidence-based methods for evaluating AI as capabilities continue to advance. He said the initiative is intended to provide a structured approach for assessing risk while helping utilities deploy AI in ways that strengthen reliable and secure grid operations.

Founding members of SAFERai.power include Alliant, Ameren, American Transmission Co. (ATC), Arizona Public Service, CenterPoint Energy, Con Edison, Constellation, Cooperative Energy, CPS Energy, Duke Energy, Exelon, HData, Microsoft, Midcontinent Independent System Operator (MISO), NERC, Nebraska Public Power District, New York Power Authority, NVIDIA, Omaha Public Power District, Pacific Gas & Electric, PJM Interconnection, Salt River Project, SHI International, Southern California Edison, Southern Company, Tennessee Valley Authority (TVA), WEC Energy Group and World Wide Technology (WWT).