What happened
Anthropic CEO Dario Amodei published an essay called "We Must Pace the Frontier". His argument: AI could bring enormous benefits, but the risks are serious, and a race to the bottom driven by commercial incentives makes them worse.
He lists losing control of AI systems, misuse for cyberattacks and bioterrorism, and serious economic disruption.
The ask, per the BBC: slow the pace of model development and monitor it closely, so companies and governments get time to deal with the risks. Building AI at all is not in question.
Then the rivals showed up. OpenAI's Sam Altman and Elon Musk said they agree with Amodei, and TechCrunch reports Google's Demis Hassabis backed the call too. Altman said OpenAI would join Anthropic in embedding third-party evaluators inside the company to monitor for safety.
On Tuesday, OpenAI global policy chief Chris Lehane told reporters that OpenAI has been working with Anthropic and Google DeepMind on AI safety for weeks, as Bloomberg first reported. Lehane is reportedly in Washington working with U.S. lawmakers on catastrophic risks.
The three labs were already talking before the essay dropped.
Why it matters
Frontier labs usually agree on two things: their model is the best one, and the other lab's model is a bit suspicious. A shared position on slowing down is unusual, and it points toward the industry trying to regulate itself.
There is a catch. Altman and others have noted that this kind of coordination could risk violating antitrust law if it is found to suppress competition. Amodei's essay proposes a narrow government waiver that would allow safety coordination.
The BBC piece also carries a darker note. A researcher who left Anthropic, Jacob Coxon, told the BBC that people at AI companies are "genuinely frightened". That is his claim, not a consensus, but it sets the mood.
Also worth noting: a pledge is not a pause. Nobody has announced a slowdown schedule, and the facts so far are an essay, endorsements, talks and an evaluator commitment. Judge the follow-through, not the vibes.
What to do with it
- Watch the evaluators.Look for who the third-party evaluators are and what access they actually get at OpenAI and Anthropic.
- Follow the waiver.Track whether lawmakers take up the antitrust waiver Amodei proposed, since it decides whether the labs can coordinate legally.
- Read the essay itself.It is public on Amodei's site, so form your own view instead of relying on hot takes.
- Don't plan around a pause.Keep building and budgeting as if the models keep shipping until someone announces otherwise.
Ciao, sapiens. When the competitors start sharing notes, pay attention to the notes.

