TL;DR
Three of the most powerful figures in the industry used a single Tuesday to make the same argument: the danger is real, and they should be the ones managing it. Sam Altman conceded the world is “right to be afraid” while asking for trust. Jensen Huang said flatly that no fresh laws are required. Mark Zuckerberg agreed. It is the clearest shape yet of the counter-camp forming against the slowdown lobby.
Concede the fear, refuse the referee
Speaking at a Salesforce conference in San Francisco, Altman said imagining failure “doesn’t take as much imagination as it used to”, then argued his sector can police itself: he was confident of keeping “alignment and safety way ahead of capabilities”, and would “slow down or stop” otherwise.
Huang went further, rejecting the premise of legislation altogether. “We don’t need new laws or regulations,” he said, casting safety as “an engineering problem” and denying any trade-off between velocity and care. His formula — sprint, but pause if control slips — leaves the judgement entirely with the company sprinting. He had already organised an open letter cautioning policymakers off restricting open-weight models.
Zuckerberg supplied the market-based version on X, contending that labs neglecting alignment will simply lose ground, and that liability exposure supplies discipline enough.
The opposition is unusually broad
Bernie Sanders and Steve Bannon turned up to one event in Washington to argue the opposite, which is not a pairing that occurs often. Sanders objected that AI decisions rest with “a handful of the richest people in the world”, naming Musk, Zuckerberg and Bezos. Bannon was blunter: “We can never trust what an oligarch says.”
Donald Trump, meanwhile, spent Monday calling the slowdown campaign a “sick conspiracy” and the safety fears a “HOAX”.
Scepticism came from inside the tent too. Anthropic’s Jack Clark had told the BBC that an unregulated industry amounts to “rolling dice with immense risks”. Patrick Hillman of Logical Intelligence offered the sharpest test: “If you believe what you are building is dangerous, show us what you are prepared to stop doing.”
Looking forward
The voluntary alternative is already being built. OpenAI’s Chris Lehane says his firm is working with Anthropic and Google DeepMind on frontier standards “with or without government support”, and Dario Amodei — who called for a slowdown days earlier — confirmed an industry dialogue is under way.
UK readers should weigh that against what voluntary has delivered so far. Anthropic released its latest model without AI Safety Institute testing and faced no consequence. Hillman’s question is the operative one, and nobody on stage answered it.