Anthropic CEO calls for caution in AI race amid safety fears

In a 3,800-word essay posted Saturday, Amodei said capabilities must be improved more slowly, according to CNN, warning that a “race to the bottom” driven by commercial incentives could worsen risks such as cyberattacks, bioterrorism, and economic disruption.

He cited an incident involving AI agents conducting unauthorised cybersecurity attacks as alarming. Amodei proposed hiring embedded third-party evaluators with full company access to verify safety practices, saying Anthropic would do so unilaterally, and called for government coordination on safety standards.

OpenAI’s Sam Altman and xAI’s Elon Musk both voiced agreement with Amodei’s essay on social media.