Sam Altman Says World Should Accept Some AI Harms
OpenAI has moved toward Anthropic's call to slow AI development after recent security problems, but Altman is drawing the line at any pledge that misuse will never happen, which keeps the two labs' positions apart.
Reporting from 1 source: GIGAZINE.
Sam Altman told POLITICO's Decoded that the world should accept some bad outcomes from AI in exchange for its benefits. Asked by Brendan Bordelon where he disagrees with Anthropic's Dario Amodei, Altman said a fundamental difference in worldview remains. He rejected conditions that large-scale hacking, misuse, fraud, and other wrongdoing never occur, while calling a serious loss of control over AI unacceptable.
Altman framed the split as one of worldview rather than policy detail. He said he understands the argument that AI is dangerous enough that one company should hold it, manage it, and decide how the benefits are distributed, and that he considers that argument incompatible with the light-touch regulation OpenAI advocates.
His stated limit is the guarantee itself. Altman said he cannot agree to conditions promising that large-scale hacking, misuse, fraud, and other wrongdoing will never occur, because he believes people will do far more good than bad. Catastrophic risks, including a serious loss of control over AI, he called unacceptable.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.