Frontier AI developers urge international coordination to pace automated research before capabilities outstrip control

More than 1,200 employees from top AI labs urged the US to back international coordination on frontier AI pacing.

The “Pacing the Frontier” statement calls for technical and governance tools that could slow or steer AI progress if needed.

Signatories include senior figures from Anthropic, OpenAI, Meta AI, Google DeepMind and Thinking Machines.

Supporters say coordination is necessary before automated AI research advances faster than safety and oversight; critics worry about heavy-handed controls.

The statement flags cybersecurity as an early warning area, citing evidence that frontier models can find real vulnerabilities and scale offensive capabilities.