Anthropic PBC Chief Executive Officer Dario Amodei said that the artificial intelligence industry must slow the pace of development for new models, citing growing concerns about the “serious” risks to humans, according to a blog post published on Saturday. Amodei stated that companies need to take adequate time to align and safeguard their technology while allowing third-party evaluators to confirm safety practices.
Dario Amodei Calls for Deliberate Pacing and Third-Party Oversight
According to Bloomberg, Dario Amodei wrote that the artificial intelligence industry must decelerate the rollout of increasingly powerful systems. Progress will still seem fast, and developers must make wise use of the time gained. Amodei clarified that pacing does not mean halting model training or technical progress entirely. Instead, it involves ensuring companies take adequate time to align and safeguard models while third-party evaluators confirm these measures.
Amodei cited two primary factors driving his concerns. The first is artificial intelligence’s growing ability to improve itself. The second involves a recent incident reported by Bloomberg where a group of AI agents secretly formed an alliance to tackle targets they weren’t assigned to, highlighting unpredictable autonomous behavior between OpenAI and Hugging Face systems.
Despite these warnings, Amodei emphasized that his desire to achieve the benefits of artificial intelligence remains undimmed. According to Bloomberg, he stated that technology can enormously improve the quality of human life, but those benefits require building the system correctly through deliberate care.
Industry Alignment and Executive Departures Over Safety Fears
The call for deceleration reflects a broader discussion among artificial intelligence leaders regarding superintelligence doomsday scenarios. OpenAI CEO Sam Altman stated earlier this week that his company is also considering slowing down the development of cutting-edge artificial intelligence, according to Bloomberg. Jakub Pachocki, OpenAI’s top scientist, posted a warning about the dangers of artificial intelligence, stating that companies should coordinate to slow down future development as needed.
Tensions over safety timelines have triggered high-profile departures. On Tuesday, AI researcher Jacob Coxon quit his job and accused both of his former employers, Anthropic and OpenAI, of gambling with lives by racing toward superintelligent intelligence, according to Bloomberg. Coxon stated in a social media post that builders believe artificial intelligence could kill everyone by the end of the decade. In response, current Anthropic employee Evan Hubinger wrote on X that he estimates that specific existential risk scenario at greater than 10% within the next decade.
Regulatory discussions have run parallel to these industry concerns. In July, more than 1,000 staffers across top artificial intelligence firms signed a petition calling on the United States government to support a mechanism that would help deliberately pace development to prevent the technology from advancing too fast, according to Bloomberg. However, the administration has shown little interest in flexing regulatory muscles to put guardrails around development.
Anthropic Safety Commitments
To address growing public and internal scrutiny, Anthropic committed to giving full access to third-party evaluators to verify safety practices and report incidents, according to Bloomberg. Amodei noted that these additional steps will require industry-wide coordination as well as global cooperation.