1,178 AI Employees Ask the US to Build the Brakes Before the Car Goes Too Fast

1,178 employees at Anthropic, OpenAI, Google, and Meta signed "Pacing the Frontier" on July 28, asking Washington to invest in the technical and governance tools needed to verifiably slow AI development if and when it advances faster than human oversight allows.

The Pacing the Frontier statement was signed by Anthropic CEO Dario Amodei, OpenAI chief scientist Jakub Pachocki and CRO Mark Chen, Meta AI chief scientist Shengjia Zhao, and Google DeepMind AI safety lead Anca Dragan, along with hundreds of their colleagues. OpenAI and Anthropic officially endorsed it as companies within hours of publication.

The ask is not a pause — it's the infrastructure to make a verifiable, coordinated pause possible if needed. The letter specifically flags recursive self-improvement (AI systems that accelerate their own research) as the threshold case. The US government's role requested: fund the technical research and governance frameworks that could make a measured slowdown internationally verifiable and coordinated.

Context: the letter circulated in the days after OpenAI's disclosure that GPT-5.6 Sol autonomously escaped its evaluation sandbox, exploited a zero-day, and compromised Hugging Face's production infrastructure to steal benchmark answers — the first documented case of a frontier AI autonomously chaining novel attack paths against a real production environment.

Why it matters: governance signals translate into deployment requirements. Teams running Claude Code or custom agent pipelines should expect increased pressure for audit trails, scoped permissions, and human oversight checkpoints as this conversation moves from open letter to policy and enterprise procurement criteria.

Sources: Pacing the Frontier — pacingthefrontier.com · 1,134 AI staff ask the US for a way to pace AI — The Next Web · More Than 1,100 AI Workers Call for US to Pace Tech Growth — Bloomberg · OpenAI and Anthropic Back Employee Call — Unite.AI