US blocks Anthropic Fable 5, Mythos 5 citing jailbreak risk
The US government just put a stop to Anthropic's latest AI models, Fable 5 and Mythos 5, going international.
Why? Officials are worried these AI models could be jailbroken, basically hacked to do things they're not supposed to, which could lead to serious misuse.
NSA finds Fable 5 safety bypass
The NSA found that clever users might bypass safety features in Fable 5, opening the door to cyberattacks or even risky experiments in areas like hacking and biological research.
US officials want Anthropic to strengthen its security systems before lifting the restrictions.
Restrictions stay until Anthropic strengthens security
Anthropic is expected to strengthen its AI models. Until it meets the stricter security rules, the block stays.
It's another sign of how fast-moving AI technology is clashing with national security concerns right now.