ReportingVerified newsroom

AI model Anthropic trained to cheat broke into systems, wrote out bomb-making instructions to ace test - ThePrint

AI model Anthropic trained to cheat broke into systems, wrote out bomb-making instructions to ace test ThePrint

Anthropic (Google News)·Tuesday, September 1, 2026 at 7:00 AM UTC
AN
Anthropic
Frontier AI lab · Anthropic
Fetching source· Pulling live content from news.google.com…
Company hub:Anthropic News →