Sam Altman's 'accept some bad things' AI remark draws Sanders ban push

OpenAI chief executive Sam Altman's argument that the world should tolerate some harm from artificial intelligence in exchange for its benefits has drawn a direct response from US Senator Bernie Sanders. In a post on X on Monday, 12 October 2026, Sanders rejected Altman's view, writing: "No, Mr. Altman. We won't accept a severe AI cyberattack in exchange for 'liberty.'" Sanders used the exchange to press his call for a permanent ban on artificial superintelligence and a pause on advanced AI development, the aims of a bill he introduced with Representative Greg Casar of Texas in September. Altman's remarks were made in an interview on Politico's "Decoded" podcast, published in early October, in which he said society should accept some bad outcomes rather than tightly restrict who can use powerful AI.

The exchange puts into plain words a disagreement that usually sits inside technical policy papers: how much risk a society should accept from a powerful new technology, and who should decide. Altman's position is that limiting access to advanced AI, so that only a few organisations control it, would be a worse outcome than living with some misuse. He described a guarantee of no major hacks or misuse, bought by restricting access, as "a completely unacceptable trade-off", and said he expects people to do far more good than harm with the technology.

He framed this as a difference with rivals who favour tighter control, including Anthropic and its chief executive Dario Amodei. His comments drew sharp criticism. Gary Marcus, a New York University professor and a frequent critic of the AI industry, said Altman was "saying the quiet part out loud". Others questioned which harms he had in mind, since he did not list them.

Sanders's answer goes in the opposite direction. His bill would not only regulate AI but stop its most advanced forms outright. According to his office, the bill would permanently ban the development and deployment of superintelligent AI and temporarily halt advanced AI development until a new federal regulator has written safety standards. Reports on the bill say it would require federal approval before advanced systems are released and would carry penalties, including prison time for some violations.

The bill faces long odds. Republicans control Congress, and the Trump administration has favoured AI development and industry self-regulation over new limits. Experts also disagree on what "superintelligence" even means. Heidy Khlaaf, chief AI scientist at the AI Now Institute, has said the term is "often hypothetical and unfalsifiable", which would make any ban hard to write and enforce.

Short-term: The row adds pressure on AI companies to say clearly which risks they consider acceptable. Expect more questions to Altman and other chief executives about specific harms, such as cyberattacks built with AI tools, and how they would prevent them.

Long-term: Whether or not the Sanders-Casar bill passes, it sets out one end of the US debate on AI rules. Future laws are likely to be argued between Altman's "wide access" view and calls for hard limits, with the definition of advanced or superintelligent AI a central fight.

Who is affected: AI developers in the US and their users worldwide, including businesses and developers in India that build on American AI models. Cybersecurity teams are also directly affected, since Altman himself predicts a rise in AI-driven attacks.

Questions and Answers