Anthropic CEO Calls for Slowing AI Development to Ensure Safety Checks

Anthropic CEO Dario Amodei warned that artificial intelligence could potentially lead a swarm to take over the internet within six to 12 months without slowed development.

Amodei proposed that frontier AI companies grant outside evaluators ongoing employee-like access to monitor safety practices and suggested U.S. government antitrust waivers.

The statements followed recent resignations by safety researchers and prior disclosures regarding blocked malicious attempts and cybersecurity incidents.