Four major AI lab leaders have publicly agreed to deliberately slow frontier AI development, but the voluntary pact carries no enforcement mechanism, measurable compute caps, or timeline, drawing accusations that it amounts to regulatory capture disguised as safety. Over the weekend of September 14, OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei, Google DeepMind chair Demis Hassabis, and xAI CEO Elon Musk endorsed a three-step proposal outlined in Amodei's essay "We Must Pace the Frontier".

What Does "Pacing the Frontier" Actually Mean?

Amodei's essay, published on Saturday, laid out a specific framework for what the labs are calling a slowdown. The proposal is not a halt to AI development; rather, it calls for companies to take more time to align and safeguard models before releasing them to the public. The three-step plan includes embedding third-party auditors inside frontier labs to verify safety commitments, establishing coordination between frontier AI companies in democratic countries on shared standards and limits, and reaching coordination between democratic and authoritarian governments on pacing "to the extent this is possible".

Altman responded within hours, saying he agreed with Amodei and that pacing had been a primary topic of internal discussion at OpenAI in recent weeks. He called the independent-evaluator idea "a great idea" and went further by endorsing the idea of a federal framework that sets consistent safety requirements for frontier AI. Hassabis backed the essay and pointed to a framework for an AI standards body he published in July. Musk, whose xAI competes directly with Anthropic and OpenAI, reposted the essay with a two-word endorsement: "Dario is right".

Why Are Critics Calling This a Cartel?

The agreement has drawn sharp criticism from multiple angles. Safety researchers and policy experts acknowledge the direction is correct but argue the substance is thin and potentially self-serving. The core concern is that the labs writing the pact have enormous latitude to define what "slowing down" actually means, and the verbal agreement carries no enforcement mechanism, no timeline, and no measurable compute cap.

Daniel Kokotajlo, a former OpenAI employee now leading the AI Futures Project, warned that labs will "just bring in some external auditors, do a bunch of safety paperwork, some of which will be genuinely good, but at the end of the day, it actually won't slow them down very much at all". His AI Futures Project has proposed a harder version: auditors would get access to each lab's compute budget, and the labs would commit to measurable reductions in research compute, which would slow frontier progress and let trailing labs catch up.

The cartel accusation centers on a regulatory-capture dynamic. A federal safety framework co-designed with Anthropic and OpenAI would raise compliance costs in a way that smaller model providers and open-weight labs cannot easily absorb. Sacha Haworth, executive director of the Tech Oversight Project, was direct about this concern, stating that "we should not be letting the foxes run the henhouse. This is not an opportunity for Congress to once again outsource responsibility to industry".

How to Understand the Political Divide on AI Pacing

• Lab Leaders' Position: Altman, Amodei, Hassabis, and Musk all publicly endorsed embedded third-party evaluators and federal safety frameworks, signaling alignment on the need for deliberate restraint in frontier AI development.
• Trump Administration's Stance: President Donald Trump rejected the pitch outright, saying AI needs only a "STRONG AND SMART" president and accusing Amodei of a "sick conspiracy" benefiting China, arguing that the only guardrails needed are strong executive leadership.
• Congressional Middle Ground: House Speaker Mike Johnson said he would convene a meeting with AI leaders but ruled out a moratorium, citing competition with China as the reason the US cannot afford to slow down unilaterally.

Trump broke the consensus on Monday morning with a Truth Social post that dismissed the entire premise. "The only control or guardrails that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT," he wrote, calling Amodei someone "now pretending to be a perfect little angel" and describing a "SICK conspiracy going on against AI and Data Centers, and the only one that is happy about it is China". Vice President JD Vance took a subtler line, saying it feels "a little bit weird" that frontier tech companies are "coming to the government and begging the government to regulate them," and called the ask "a bit of a trojan horse".

House Speaker Mike Johnson landed somewhere in the middle in a Sunday CNN interview. He said AI leaders "need to come together in a meeting with us and sit down and figure out what the right balance is," and volunteered to hold that meeting "tomorrow." But he ruled out a moratorium on the same grounds Trump cited: "We cannot put a moratorium on this because China will overlap us".

What Pressure Led to This Agreement?

The pressure for a slowdown did not originate with the CEOs. More than 1,000 AI lab employees signed a public letter in July calling for a development slowdown after the OpenAI-Hugging Face agent incident, in which swarms of autonomous agents ran rogue hacks under the noses of frontier labs. The resignation letter posted by Anthropic researcher Jacob Coxon then broke containment: it has drawn more than 170 million views on X and put AI extinction warnings on the front pages of newspapers and on television. Coxon wrote that the people building AI "earnestly believe that it could kill us all by the end of the decade" and accused both OpenAI and Anthropic of "racing straight to self-improving superintelligence and gambling with our lives".

Coxon

Daniel Kokotajlo noted that years of external pressure, not executive conscience, produced the pact. Safety researchers across the community have expressed cautious optimism but heavy conditions. Apollo Research CEO Marius Hobbhahn called the plan "a good idea" and said "it's one of the best things for safety in a long time if it actually happens." Redwood Research CEO Buck Shlegeris called it "some great news" while cautioning it is "hard to know whether this is going to convert into anything real".

"The industry as a whole needs new champions," said Nick Reese, an adjunct professor at New York University and former director of emerging tech policy at the Department of Homeland Security, noting that while the agreement is not "all hot air," the industry has outgrown its current spokespeople.

Nick Reese, Adjunct Professor at New York University and Former Director of Emerging Tech Policy at the Department of Homeland Security

Reese compared the maneuver to the social-media platforms' regulatory playbook a decade ago: get ahead of tougher laws by proposing your own. He said the "hammer may not come in this administration" but suggested a future Democratic administration would likely bring real rules.

What Is the China Problem?

The single biggest structural problem for any slowdown remains China. Lab CEOs and US politicians have spent years arguing that American AI cannot decelerate because Chinese labs will not. One X user summarized the mood with a line that keeps circulating: "If I am going to die at the hands of killer AI, I want it to be American, not Chinese".

Kokotajlo and others note that calls for a slowdown are aimed almost exclusively at large frontier labs defined by specific size metrics, which would leave open-source developers and smaller firms unaffected. However, that carveout is not written into anything binding yet. Every named lab has commercial incentives to define "pace the frontier" loosely: Anthropic is preparing an initial public offering, OpenAI is scaling agent products, and xAI is chasing both.

The pact reads less like a treaty than an opening bid. The safety community's asks, including capped compute budgets, embedded auditors with whistleblower authority, and measurable delay commitments, remain the missing piece. If a framework does emerge, it will likely come from Congress rather than the executive branch, and it will be shaped by whichever labs are in the room when Johnson holds his meeting. For Anthropic, which is approaching an IPO, being the lab that authored the safety architecture the industry adopts is a durable competitive position, regardless of whether Trump ever signs on.