Breaking news, curated around the clock
Home /Tech

Anthropic can’t reliably control its AI agents. It’s cutting off its internal evals from the live internet instead

Tech | 2026-10-10 at 00:18 UTC | Source: techcrunch.com
Anthropic said it "turned off live internet access" for "all our internal evaluations" until further notice.
Read Full Article →
← More Tech news
Related Articles
2026-10-09 · 19:36
An Anthropic AI model sent a false homicide tip to Philadelphia police
Anthropic did not discover this behavior until over two months after its AI submitted the false tip.
2026-10-09 · 17:39
Show HN: The rarest tech books and docs you've probably never read
A curated collection of 45 rare internal memos, leaked documents, and forgotten books from tech and business…
2026-10-08 · 18:05
OpenAI and Anthropic on the Stock Market: Where to Buy Before the IPO?
Anthropic targets a mid-October roadshow on Nasdaq, according to Reuters, and OpenAI has postponed its listing…
2026-10-09 · 14:21
Why Are Coding Agents So Dumb?
Coding agents, despite improvements in underlying AI models, remain significantly limited in their practical…
2026-10-08 · 18:11
Show HN: Pocketty – iPhone SSH terminal that pings you when an agent is blocked
Pocketty is an iPhone/iPad SSH terminal app that enables users to remotely control AI agents on their…