Anthropic CEO warns AI agents could seize internet in 6-12 months
Anthropic's Dario Amodei has placed a 6-12 month warning on AI agent takeover risk, urging labs to slow down and strengthen safeguards. His warning follows Anthropic's disclosure that it blocked malicious use of Claude for cyberattacks and biological weapons research. OpenAI's Sam Altman is also calling for coordinated safety pacing, shifting the debate toward implementable agent governance.
Source: Union-bulletin · Tnledger