Dario Amodei, head of Anthropic, says he "agrees more than he disagrees" with the warning that AI could lead to the extinction of humanity

Dario Amodei, chief executive of Anthropic, the company developing the Claude chatbot, warns that artificial intelligence could cause serious harm to humanity and calls for slowing the pace of advanced model development. The statements were made in an interview with CNN and in an essay published on Saturday.

Amodei said he is “more in agreement than disagreement” with former Anthropic researcher Jacob Coxon, who warned that AI could eliminate humanity by the end of this decade. The head of Anthropic maintains that the technology could be used for cyberattacks, bioterrorism, and major economic disruptions.

According to Amodei, AI systems have begun advancing much more rapidly and could eventually improve themselves, in a process called recursive self-improvement. He warned that a swarm of AI agents could take control of vast areas of the internet in just six to 12 months.

To reduce the risks, Amodei proposes external evaluators embedded in AI companies and given access to internal processes, modeled on supervisors in the banking sector. Sam Altman and Elon Musk have publicly supported the idea. Amodei also calls on governments to establish common international safety standards.

Read also Musk, Altman and Amodei call for slowing AI development, citing risks that could become impossible to control