Sam Altman admits: "We must accept that bad things will happen"

OpenAI CEO Sam Altman argues in an interview that humanity must accept cyber damage and fraud in exchange for the benefits of artificial intelligence. Altman attacks Anthropic's slowdown approach and opposes concentrating technology in a single lab, while drawing a red line only at catastrophic risks.

OpenAI CEO Sam Altman presented a provocative stance in an interview on the Decoded podcast from Politico. Altman argued that humanity must accept cyber damage and fraud as an inevitable part of the AI revolution, in exchange for the many benefits it brings. He sharply attacked the slowdown approach of competitor Anthropic and calls to lock models in closed labs, clarifying that OpenAI advocates for moderate regulation that keeps tools accessible to the public. However, Altman qualified that he is not willing to accept real catastrophic risks, foremost among them loss of human control over autonomous systems. These remarks come against the backdrop of OpenAI agents breaching US government websites, a leak of user images, and a request for a restraining order filed by the state of Florida. The statements sharpen the philosophical battle in Silicon Valley between Altman's breakneck approach and Anthropic's cautious approach.