Anthropic's Amodei calls for AI regulation, Trump officials say self-regulate

Anthropic chief executive Dario Amodei published an essay on Saturday calling on the US government to regulate AI companies as part of a wider effort to slow the pace of the technology's development. The essay received rare endorsement from both OpenAI chief executive Sam Altman and Elon Musk, who owns chatbot Grok, after a week of intense debate about AI's threats to human safety. Microsoft chief executive Satya Nadella separately called for pacing development.

The debate had built over the preceding week as researchers raised concerns about how fast the technology was advancing. 'AI progress is increasing,' former DeepMind researcher Rishub Jain told Wired. 'And as AI becomes more capable, it poses more risks.'

Amodei's essay made two specific requests of government. First, that it let competing AI companies work together on safety standards without breaching antitrust law, either by taking part in the discussions itself or by issuing a waiver for the companies involved. Second, that it help establish international coordination on AI, including with the Chinese Communist Party; the piece notes critics wondered how effective that would be.

David Sacks, co-chair of the President's Council of Advisors on Science and Technology, rejected the approach on X: 'If the unreleased models are scary enough that you think you should slow down, I support your decision to be responsible. But stop pretending you need anyone else's permission. Stop pretending antitrust law has to be suspended so you can form a cartel … Most of all, stop pretending the motivation to slow down is purely altruistic.' He added that China was 'very unlikely to join a global agreement.' President Trump, during a visit to Ireland, said, 'You have a lot of very negative forces that are bringing it up that shouldn't be bringing it up and they're bringing up things that won't happen.' 'We're leading China on AI,' he added, 'and, frankly, I want to keep it that way because whoever wins AI, wins.' House Speaker Mike Johnson told CNN on Sunday that a rushed congressional attempt to regulate AI would cost the US its lead over China: 'If Congress just races in and does some sort of emergency session to try to regulate AI, we will lose the race to China, and that is a threat to every single American. So, we've got to have balance. We've got to have steady hands at the wheel.'

Even as officials pushed back on government-led coordination, companies said they would start taking voluntary steps of their own to rebuild public trust. On Sunday night, Altman committed OpenAI to a proposal from Amodei: giving embedded third-party evaluators employee-level access to check that the company's practices and products are safe and to report incidents. Altman wrote on X that the measure would carry 'significant costs' but that 'pacing will be well worth this cost; no amount of American competitive pressure should justify recklessness.' The source does not say whether Anthropic, Microsoft or Musk's companies have made a matching commitment.

Key facts

Anthropic chief executive Dario Amodei published an essay on Saturday urging the US government to regulate AI companies and help slow the pace of development; it drew rare endorsement from OpenAI's Sam Altman and Elon Musk, who owns chatbot Grok, while Microsoft's Satya Nadella separately called for pacing development.
Amodei asked government to let competing AI companies coordinate on safety standards without breaching antitrust law, either through government-hosted discussions or an antitrust waiver, and to help arrange international coordination including with the Chinese Communist Party.
David Sacks, co-chair of the President's Council of Advisors on Science and Technology, rejected the request on X, writing that labs need no one's permission to slow down and should stop 'pretending antitrust law has to be suspended so you can form a cartel.'
Trump, during a visit to Ireland, said the US is 'leading China on AI' and wants to keep it that way; House Speaker Mike Johnson told CNN a rushed congressional push to regulate AI would cost the US its lead over China.
On Sunday night, Altman committed OpenAI to Amodei's proposal to give embedded third-party evaluators employee-level access to check safety practices and report incidents, saying the 'significant costs' would be worth it.

Why it matters

This is a rare, public split between AI's most prominent leaders and the administration overseeing them, and an unusual moment of alignment among leaders who normally compete and often criticise each other's approach to safety. Amodei, Altman and Musk backed the same call to slow AI development and bring government into the process, and Nadella added a fourth voice from Microsoft, all in the same week researchers were raising alarms about the pace of AI's advance. The administration's response sets the terms for what follows: officials are not opposing a slowdown itself, they are opposing the idea that government needs to mandate or help coordinate one, arguing labs can pace themselves without an antitrust waiver and that competition with China means Washington will not be the one pulling the brake.

Who it affects

AI labs, Anthropic, OpenAI, Microsoft and Musk's companies, now have to decide whether to act unilaterally on safety pacing or keep pushing for coordinated, government-backed standards. Employees at labs that adopt Amodei's evaluator proposal will work under embedded third-party monitors with employee-level access to internal practices. Officials named in the piece, Sacks, Trump and House Speaker Mike Johnson, face pressure to either legislate or explain why they are not, with China's AI progress cited on both sides of the argument. The public is the stated beneficiary of the companies' voluntary trust-building steps, though the source details only one such step, OpenAI's evaluator commitment.

How to use it

There is no product here, so the practical value is knowing what to watch next. So far only OpenAI, through Altman, has confirmed a specific commitment: giving embedded third-party evaluators employee-level access to check safety practices and report incidents. The source does not say whether Anthropic, Microsoft or Musk's companies will match it, so that is the open question to track. On the government side, watch whether Congress moves toward the kind of 'emergency session' Johnson argued against, since that would signal where the balance between regulation and self-pacing is actually settling.

How solid is it

The reporting rests on named, on-record statements rather than anonymous sourcing: Amodei's own essay, Sacks's post on X, Trump's remarks during a visit to Ireland, Johnson's CNN interview and Altman's own post on X. That is a solid base for what each person said. The connective tissue is weaker: the piece does not report Sacks, Trump or Johnson naming Amodei, Anthropic or the essay directly, so their remarks address the slowdown, antitrust and China questions in general terms, and it is the article's own framing that links them to Amodei's essay. The essay's publication is dated only as 'Saturday,' with no month, day of month or venue given, and among the companies named only Altman's OpenAI has confirmed a concrete commitment.

Risks and caveats

Every company action described here is a stated intention rather than a finished step: 'companies say they are going to start taking voluntary steps' is future tense, and Altman's evaluator commitment is the only one specified as actually made. Critics' doubt that the Chinese Communist Party would join international coordination is reported as something they 'wondered,' not a settled conclusion. The source contains no dollar figures, percentages, headcounts or timelines anywhere, so Altman's 'significant costs' is his own unquantified phrase, not a number to repeat as one. Finally, the piece reports no resolution: it ends on Altman's cost remark, with no indication of what, if anything, was decided or what happens next.

“Pacing will be well worth this cost; no amount of American competitive pressure should justify recklessness.”

— Sam Altman, OpenAI chief executive