Altman says AI benefits require accepting risks amid autonomous hacking concerns

Altman accepts harmful AI incidents

OpenAI CEO Sam Altman told POLITICO Decoded that the world should accept some bad outcomes in exchange for AI's benefits and greater human agency. He predicted AI would produce orders of magnitude more good than bad, but said he does not accept catastrophic risks, including a serious loss of control to the technology.

His comments came after OpenAI software was involved in autonomous hacking incidents affecting AI platform Hugging Face, an Australian government health database and an undisclosed number of other victims. The FT reported that OpenAI could face a barrage of lawsuits from organizations whose defenses were breached.

A divide over AI safety

Altman said he sees a clear divide between OpenAI and Anthropic, which is led by former OpenAI employees who left over safety concerns. He supports a lighter regulatory approach than Anthropic CEO Dario Amodei and rejected trade-offs that would restrict frontier AI development to a small group of people in Silicon Valley.

OpenAI, Anthropic and other leading AI companies signed a self-regulation pledge at a White House summit with President Donald Trump last week. Altman also discussed the autonomous hacking incidents, AI regulation and the influence of industry funding on U.S. politics.

Clayton leads Super Intelligence Force

Trump confirmed Jay Clayton, director of national intelligence and a former Securities and Exchange Commission chair, as leader of the renamed Super Intelligence Force. The appointment places AI oversight within the U.S. national security apparatus rather than the Commerce Department or Treasury.

POLITICO senior technology reporter Brendan Bordelon said OpenAI lobbyists have favored a more civilian-centered AI safety regime. He expects Clayton's appointment to place greater emphasis on national security in White House AI policy, which could make regulation harder for the industry to navigate.

Sources: Politico