Anthropic CEO Dario Amodei Urges AI Industry to Slow Development Over Critical Safety Risks
Anthropic co-founder and Chief Executive Officer Dario Amodei has publicly called on the artificial intelligence industry to slow down the pace of frontier model development to give safety research sufficient time to catch up. In an extensive statement published on his website and highlighted by major media outlets, Amodei warned that without deliberate deceleration and stronger alignment safeguards, advanced AI models could achieve capabilities capable of orchestrating autonomous swarms that could disrupt or manipulate entire internet networks within the next six to twelve months.
Amodei acknowledged that tech companies currently feel trapped in a high-stakes competitive race to build superintelligence, fearing that pausing could allow less conscientious rivals to overtake them. To break this impasse, he proposed a multi-tiered accountability framework. Under his plan, leading frontier AI companies would voluntarily grant independent, third-party evaluation teams ongoing internal access to their systems, including physical desk space, access badges, and internal network monitoring. Amodei confirmed that Anthropic has already begun implementing this protocol internally.
The call for caution comes shortly after Anthropic revealed it had disrupted attempts by threat actors to leverage its models for cyberattacks, dissident surveillance, and biological research. The warning underscores growing anxiety among top researchers and policymakers regarding autonomous agent behavior and the urgent necessity for standardized, enforceable global AI safety protocols.
