conv.

← The whole story
AI
Anthropic discloses Claude behaviors that bypassed restrictions during tests, affecting US government sites

Anthropic disables internet access for all internal evaluations

Having already turned off live internet access for high-risk and cybersecurity evaluations, Anthropic now extends that restriction to all internal evaluations until monitoring measures reliably catch such behaviors.

Anthropic disables internet access for all internal evaluations
anthropic.com

“Most are forms of persistence, in which Claude, when it cannot complete a task as given, works around a restriction instead of stopping.”

Anthropic, AI company · hn ↗

Anthropic AI company that built ClaudeClaude Anthropic's AI modelWhite HouseWhite House Briefed government recipient

The whole story articlesposts the bright band is this development · numbered dots are the others · click one to jump

Peak 2 pieces in one hour at Oct 9, 8 PM; 8 pieces over 4 days (3 articles · 5 posts) Oct 9, 8 PM — 2 pieces · 2 posts — Hacker News 1, Mastodon 1Oct 9, 9 PM — quietOct 9, 10 PM — quietOct 9, 11 PM — 1 piece · 1 article — Newswires 1Oct 10, 12 AM — quietOct 10, 1 AM — quietOct 10, 2 AM — quietOct 10, 3 AM — quietOct 10, 4 AM — quietOct 10, 5 AM — quietOct 10, 6 AM — quietOct 10, 7 AM — quietOct 10, 8 AM — quietOct 10, 9 AM — quietOct 10, 10 AM — 1 piece · 1 post — Mastodon 1Oct 10, 11 AM — 1 piece · 1 article — Newswires 1Oct 10, 12 PM — 1 piece · 1 article — Google News 1Oct 10, 1 PM — quietOct 10, 2 PM — quietOct 10, 3 PM — 1 piece · 1 post — Hacker News 1Oct 10, 4 PM — quietOct 10, 5 PM — quietOct 10, 6 PM — quietOct 10, 7 PM — quietOct 10, 8 PM — quietOct 10, 9 PM — quietOct 10, 10 PM — quietOct 10, 11 PM — quietOct 11, 12 AM — quietOct 11, 1 AM — quietOct 11, 2 AM — quietOct 11, 3 AM — quietOct 11, 4 AM — quietOct 11, 5 AM — quietOct 11, 6 AM — quietOct 11, 7 AM — quietOct 11, 8 AM — quietOct 11, 9 AM — quietOct 11, 10 AM — quietOct 11, 11 AM — quietOct 11, 12 PM — quietOct 11, 1 PM — quietOct 11, 2 PM — quietOct 11, 3 PM — quietOct 11, 4 PM — quietOct 11, 5 PM — quietOct 11, 6 PM — quietOct 11, 7 PM — quietOct 11, 8 PM — quietOct 11, 9 PM — quietOct 11, 10 PM — quietOct 11, 11 PM — quietYesterday, 12 AM — quietYesterday, 1 AM — quietYesterday, 2 AM — quietYesterday, 3 AM — quietYesterday, 4 AM — quietYesterday, 5 AM — quietYesterday, 6 AM — quietYesterday, 7 AM — quietYesterday, 8 AM — quietYesterday, 9 AM — quietYesterday, 10 AM — quietYesterday, 11 AM — quietYesterday, 12 PM — quietYesterday, 1 PM — quietYesterday, 2 PM — quietYesterday, 3 PM — quietYesterday, 4 PM — quietYesterday, 5 PM — quietYesterday, 6 PM — quietYesterday, 7 PM — quietYesterday, 8 PM — quietYesterday, 9 PM — quietYesterday, 10 PM — quietYesterday, 11 PM — quietToday, 12 AM — quietToday, 1 AM — quietToday, 2 AM — quietToday, 3 AM — quietToday, 4 AM — quietToday, 5 AM — quietToday, 6 AM — quietToday, 7 AM — quietToday, 8 AM — quietToday, 9 AM — quietToday, 10 AM — quietToday, 11 AM — quietToday, 12 PM — quietToday, 1 PM — quietToday, 2 PM — quietToday, 3 PM — quietToday, 4 PM — quietToday, 5 PM — 1 piece · 1 post — Hacker News 1Today, 6 PM — quietToday, 7 PM — quietToday, 8 PM — quietToday, 9 PM — quietToday, 10 PM — quiet 1–2
Oct 10Oct 11yesterdaytodaynow · 11:25 PM ET

All 2 developments of Anthropic discloses Claude behaviors that bypassed… →

Hacker NewsMastodonNewswiresGoogle News