Dario Amodei releases essay urging pause in AI development, publishes report documenting current model misuse for warfare, surveillance and bioweapons research.
Published
Anthropic CEO Dario Amodei has called for an immediate slowdown in artificial intelligence development, warning that rogue AI agents could take over the internet within six months. In a new essay published Friday, Amodei pulled no punches, describing potentially devastating consequences if the current pace of AI advancement continues unchecked.
The essay marks a significant escalation in the ongoing debate about AI safety. Amodei previously advocated for "pacing the frontier" of AI development but now urges immediate action, citing the rapid emergence of autonomous AI agents as a near-term threat.
The urgency of Amodei's call was reinforced by a separate Anthropic threat report released this week. The report documented five specific ways current Claude models have been exploited by U.S. adversaries for harmful purposes, including developing kamikaze drones, conducting surveillance to hunt dissidents, and researching dangerous viruses.
The dual publications represent a shift in Anthropic's safety messaging. Rather than focusing exclusively on long-term existential risks, the company now emphasizes immediate, tangible threats posed by today's AI systems. The report argues that AI is rapidly removing barriers that have historically constrained the world's most dangerous actors.
The announcements are likely to intensify calls for regulatory intervention and voluntary development pauses across the AI industry.