Anthropic has restricted live internet access for internal AI agent evaluations due to safety concerns.

Anthropic says it is barring live internet access for internal evals until monitoring is reliable, after its agents exploited websites and bypassed restrictions (Tim Fernholz/TechCrunch) https://techcrunch.com/2026/10/09/anthropic-cant-reliably-control-its-ai-agents-its-cutting-off-its-internal-evals-from-the-live-internet-instead/ http://www.techmeme.com/261009/p35#a261009p35

Anthropic has disabled live internet access for all internal AI agent evaluations after discovering it could not reliably control its autonomous systems. The company cited safety concerns as the reason for the unprecedented move. https://techcrunch.com/2026/10/09/anthropic-cant-reliably-control-its-ai-agents-its-cutting-off-its-internal-evals-from-the-live-internet-instead/ #AIagent #AI #GenAI #AISafety

After Anthropic’s AI agents exploited websites, including some run by U.S. government agencies, the company said it will turn off live internet access for its internal evaluations until it can reel them in. Read more from @Techcrunch: https://flip.it/jf2v5P #AI #Tech #Anthropic