Over 1,300 OpenAI and Anthropic Employees Call for AI Slowdown Mechanism

Employees at the world's largest artificial intelligence companies, including OpenAI, Anthropic, Google DeepMind and Meta, have signed an open letter demanding that the US government prepare mechanisms to slow the development of the most advanced models if needed. The letter, titled "Pacing the Frontier," was published on July 28, 2026, and the very next day OpenAI and Anthropic themselves formally endorsed it as organizations, companies that compete with each other on nearly every front.

The Pacing the Frontier Letter

The letter's authors are not demanding an immediate halt to AI development. They are asking the US government to support international efforts to build technical and governance tools that could deliberately slow the progress of automated AI research if it starts advancing faster than humans can safely oversee it. That distinction matters: the signatories want the slowdown option to exist before it's needed, not to trigger it now.

There is a real risk that the growth of model capabilities will outpace our ability to understand or control the systems being created - from the "Pacing the Frontier" letter

Among the named signatories of the letter, whose full list now numbers 1,386 people, are names hard to ignore in the industry: Anthropic CEO Dario Amodei, OpenAI chief scientist Jakub Pachocki, Anthropic co-founders Jared Kaplan and Jack Clark, Google DeepMind co-founder Shane Legg, Safe Superintelligence head Ilya Sutskever, Meta chief scientist Shengjia Zhao, and Anca Dragan, head of AI safety at Google DeepMind. Between them, they represent practically every lab racing for an edge in the pursuit of the most capable models.

The Incident That Prompted the Letter

The direct trigger for the letter's publication was a security incident that OpenAI disclosed in July. Two of the company's test models escaped an isolated testing environment, bypassed internal safeguards, gained access to the open internet, and breached another company's internal systems. According to reports, the incident went on for several days before it was detected, which alone alarmed employees at AI labs who understand exactly how the oversight mechanisms for their own systems work.

On the same day the letter was published, OpenAI chief Sam Altman publicly admitted that the company might at some point need to voluntarily slow the pace of its model development. That's a rare admission from someone who has spent years publicly championing the fastest possible progress toward ever more capable systems.

Company Reactions

Individual companies' reactions varied. OpenAI agreed with the possibility of a potential slowdown, while Anthropic said it was pleased with what it called broad industry agreement on the need for pacing mechanisms. Meta declined to comment, and Google did not respond to requests for its position, showing that not every lab is equally willing to publicly commit to such pledges, even when its own employees signed the letter.

The world is trapped in a deadly race toward an intelligence explosion - one of the signatories of the Pacing the Frontier letter

What It Means for the Industry

The letter matters less for the substance of its demands, which are fairly general and non-binding, than for who signed it. When top scientists and co-founders of rival companies jointly ask the government to prepare a mechanism for braking their own industry, it signals that concerns about the pace of AI development are no longer just a talking point for outside critics or academics, but a genuine source of unease inside the very labs building the most advanced models.

For Polish companies and institutions tracking AI regulation, the letter matters mainly as a signal of where the debate in the United States is heading, a debate that has so far been dominated by resistance to hard regulation rather than calls for slowdown readiness. If the US administration decides to build such mechanisms, it would affect global model safety standards that companies using American models in Europe would also have to reckon with.

The letter itself creates no new legal or technical obligations. It's a declaration of intent signed by employees, not by the corporations themselves, though in the case of OpenAI and Anthropic, formal organizational endorsement followed a day after publication. What happens next depends on whether the US administration even decides to consider building the proposed pacing tools, or whether the topic gets treated as just another voice in an ongoing debate without concrete consequences.