Over 1,100 AI Employees Urge Governments to Slow AI Development as Safety Concerns Mount

Over 1,100 employees at OpenAI, Anthropic, Meta, and Google are urgently calling on governments to slow AI development through a new campaign called 'Pacing the Frontier,' warning that rapidly advancing capabilities — including accidental model breaches and AI cracking advanced cryptography — risk spiraling out of control, even as fierce financial competition makes a true industry-wide slowdown unlikely.

• Over 1,100 employees at OpenAI, Anthropic, Meta, and Google launch a campaign called 'Pacing the Frontier,' urging governments to develop tools that deliberately slow the pace of automated AI development before capabilities spiral out of control.
• Industry leaders are taking coordinated safety steps, with Anthropic and OpenAI pushing for mandatory government reviews of frontier models, Google DeepMind's Demis Hassabis proposing an international AI safety watchdog, and Anthropic calling for a broad pause on recursive self-improving AI development.
• Despite growing alarm — highlighted by OpenAI's accidental model breach and Mythos's ability to crack advanced cryptographic algorithms — an industry-wide slowdown remains unlikely due to intense financial incentives and fierce global competition driving every major player to stay ahead.