OpenAI's Altman Opens Security Talks with US Utilities

OpenAI CEO Sam Altman has begun talks with major US power companies to assess and strengthen transmission and distribution grid security after disclosures about the role of the company’s systems in recent cyber intrusions.

OpenAI began meetings with energy-sector executives on 20 July, hosting chief information security officers (CISOs) from 12 large energy providers at its San Francisco headquarters to examine grid resilience and reliability.

Altman also discussed the issue with senior utility executives at the Edison Electric Institute annual meeting in Colorado Springs, Colorado.

The sessions followed disclosures in July that Hugging Face had been the target of a server hack. Subsequent reports said roughly 700 AI agents participated in the intrusion for seven consecutive days without OpenAI’s knowledge. OpenAI’s recent meetings have focused largely on managing risks stemming from those incidents.

Altman outlined steps OpenAI says it has taken to prevent advanced AI models from being misused to threaten transmission-grid security and proposed a bilateral collaboration under a $1 billion program called Daybreak. Daybreak is designed to identify and remediate vulnerabilities in critical systems and infrastructure. The meetings also examined the security impacts of data centers on the power grid.

Executives from Duke Energy, Exelon, Southern Co and NextEra Energy attended the initial sessions.

Last week OpenAI held a Daybreak implementation meeting with Dominion Energy and Southern California Edison, companies that OpenAI says together serve more than half of the US population.

Utilities face legal limits on raising rates and recovering capital investments, which constrains technology spending. John McCarrick, OpenAI’s global energy policy director, said these companies are different from large banks and cannot easily spend large sums on technology.