Anthropic CEO warns AI could take over internet within a year if industry does not create safeguards

Anthropic CEO Dario Amodei warned that, without stronger safeguards, AI agents could be able to take over the internet in as little as six months to a year. Speaking days after former Anthropic safety researchers said existential risks weren’t getting enough attention, Amodei called for industry and governments to slow “pacing” and ensure increasingly capable models stay aligned with human commands and values. OpenAI CEO Sam Altman said companies should coordinate on AI safety rather than wait for new legislation, arguing development should be slower but not stopped. Recent disclosures from Anthropic include thwarted malicious uses and prior testing-related and hacking incidents.