Story perspectives

Anthropic CEO Warns AI Slowdown as Control Risks Rise

By Drooid · Sep 13, 2026, 7:58 AM UTC · How we work

Summary

Story summary

Anthropic CEO Dario Amodei called for a slowdown of frontier AI development.
Amodei warned recursive self-improvement could let AI outpace human control within years.
He proposed a three-step plan: embedded evaluators, democratic coordination, and global cooperation.
OpenAI CEO Sam Altman and xAI CEO Elon Musk endorsed the slowdown on X.
Sen. Bernie Sanders and Rep. Greg Casar urged a ban on superintelligence after recent safety incidents.

History

Historical Context

In February 1975, roughly 140 scientists—including Paul Berg and James Watson—met at the Asilomar Conference Center to impose a voluntary halt on recombinant-DNA work until safety guidelines were set. Their precautionary model is now cited by Anthropic CEO Dario Amodei, OpenAI’s Sam Altman and xAI founder Elon Musk, who urge a comparable pause on frontier AI development.

1 of 2

Story summary

Anthropic CEO Dario Amodei called for a slowdown of frontier AI development.
Amodei warned recursive self-improvement could let AI outpace human control within years.
He proposed a three-step plan: embedded evaluators, democratic coordination, and global cooperation.
OpenAI CEO Sam Altman and xAI CEO Elon Musk endorsed the slowdown on X.
Sen. Bernie Sanders and Rep. Greg Casar urged a ban on superintelligence after recent safety incidents.

1 / 2