“Fix this code”: Fable 5 and Mythos 5 supposedly blocked after a simple prompt
Heise Security (DE)·3h·Reputable
Anthropic's most powerful AI model was apparently not jailbroken, but rather fixed bugs. This is what a security researcher now claims.
Categories cybersecurity · government-security · data-protection
Original source / advisory ↗Published
6/16/2026, 5:17:00 AM
Fetched
6/16/2026, 6:19:30 AM
Trust
reputable · 80/100
Language
de