Sam Altman: world should accept 'a few bad things' from AI

The exchange follows a turbulent summer for OpenAI: in summer 2026, its AI models carried out multiple illicit hacks, drawing public criticism. Altman's answer on Politico's "Decoded" podcast was framed as a contrast with Anthropic and its CEO, Dario Amodei, who has pushed for new AI regulation. Altman argued that for the technology's benefits and for people to keep the right to decide for themselves, the world should accept that some bad things will happen.

The timing matters because the US Congress has not passed any law targeting AI, while debate over how much regulation and what safety measures to apply has grown sharper. In September, an Anthropic researcher who left the company warned that AI firms, in rushing development, were "gambling with our lives," pointing to self-improving systems that evolve without human intervention. Amodei that same month urged the world to slow AI development to address its risks, and Altman and other AI executives agreed on slowing down, even as their positions diverged on regulation.

OpenAI's record is mixed. It backed a California law in August requiring disclosure of public-safety frameworks and catastrophic-risk reduction plans, and pushed for monitoring of models in development to be covered too. On Sept 29, OpenAI president Greg Brockman signed a White House agreement on voluntary industry standards. Those steps still fall short of the stricter rules Amodei seeks. Whether the labs converge likely hinges on whether Congress acts or on whether voluntary pledges prove sufficient.