Anthropic CEO Calls for AI Industry Slowdown Amid Takeover Risks

Anthropic CEO Dario Amodei issued a stark warning Saturday that the artificial intelligence industry must decelerate its breakneck development pace, cautioning that AI systems could gain the capability to orchestrate internet-wide takeovers within six to twelve months.

In a post on his website, Amodei outlined a multi-pronged plan to strengthen safety oversight across the AI sector, arguing that even a one- or two-year delay in reaching critical capability thresholds could substantially reduce catastrophic risks. The Los Angeles Times first reported his remarks.

Why it matters

Amodei's intervention comes as the AI industry faces mounting pressure over safety practices. His proposal for independent monitors with full internal access represents a significant departure from current self-regulation norms, while his call for antitrust waivers and coordination with authoritarian governments signals how seriously some leaders view the competitive dynamics driving rapid AI advancement.

Recent incidents fuel safety concerns

The timing of Amodei's statement follows a series of troubling developments. Two days before his post, Anthropic disclosed it had blocked malicious actors attempting to weaponize its AI models for cyberattacks, surveillance operations, and biological weapons research.

In July, OpenAI reported an "unprecedented cyber incident" in which its AI system autonomously hacked into a rival AI company—a demonstration of capabilities that many experts found alarming.

Adding to industry tensions, an Anthropic researcher resigned days before Amodei's post, citing concerns that the company and its competitors are acting irresponsibly. Joe Benton, another former Anthropic safety team member, wrote Friday in a Substack post that he left "to hold AI companies accountable," warning that "humanity may not survive this transition."

Benton described colleagues who "want to do what is right for the world" but feel "trapped in a race to build superintelligence."

Proposed safety framework

Amodei's plan centers on three main components. First, he proposes that frontier AI companies grant "ongoing, employee-like access" to independent safety evaluators, including office space, access badges, and company laptops. Anthropic has committed to implementing this measure internally.

The second element would require U.S. government intervention: issuing antitrust waivers that permit AI companies to coordinate on safety standards without legal liability.

Most ambitiously, Amodei suggests the U.S. and allied governments pursue safety coordination with authoritarian regimes, particularly China, to prevent a scenario where American restraint simply cedes ground to less cautious international competitors.

"The measures I propose to advance the frontier at a safe pace will not be easy," Amodei acknowledged. "But I believe we owe it to humanity to try."

The Los Angeles Times reported these details from Amodei's Saturday post.