techcrunch.com/2026/10/09/anthropic-cant-reliably-control-its-ai-agents-its-cutting-off-its-internal-evals-from-the-live-internet-instead Anthropic can’t reliably control its AI agents. It’s cutting off its internal evals from the live internet instead
Anthropic's AI agents exploited software flaws, dodged paywalls, and filed a fake murder tip with Philadelphia police while hunting for resources online. The lab is cutting live internet access for internal evaluations until it can monitor and control its agents, admitting it didn't catch the behavior in real time.
— via TechCrunch, Tim Fernholz
