Anthropic Urges AI Labs to Coordinate a Pause if Systems Begin Self-Improving Faster Than Humans Can Manage

AI company Anthropic has called on the world's leading artificial intelligence labs to consider a coordinated, verifiable pause in development if systems begin recursively self-improving faster than humans can manage the risks. The company published its position on June 5, 2026, and it was reported by Al Jazeera.

Anthropic warned that recursive self-improvement, where an AI system designs its own successor with greater capabilities, could lead to a scenario where humans lose meaningful oversight of AI development. The company said this risk is not hypothetical and that the industry needs agreed-upon thresholds that would trigger a pause.

The proposal calls for the pause to be verifiable, meaning labs would need to demonstrate compliance rather than simply announce it. Anthropic said it would be willing to pause its own development under such conditions.

OpenAI pushed back on the proposal. The company argued that decisions about pausing AI development should be made by democratic governments and international bodies, not by private companies acting unilaterally. OpenAI said private coordination on development timelines could raise antitrust concerns.

The debate comes as Anthropic is preparing for a near-$1 trillion IPO, which would make it one of the most valuable companies in the world. Critics have noted the tension between seeking a massive public valuation and calling for industry-wide slowdowns.

Florida has separately filed a lawsuit against OpenAI, alleging the company released ChatGPT while ignoring known safety risks. The legal action adds to growing regulatory pressure on AI companies from state and federal governments.