Anthropic’s AI model Fable 5 has been allowed to resume global distribution after a two-week ban imposed by the US government. The suspension followed the discovery of a jailbreak exploit by Amazon researchers, which raised security concerns. According to The Decoder, the ban was lifted once Anthropic implemented a new safety classifier blocking the exploit in over 99 percent of cases.
Despite the fix, Anthropic has warned that smaller models, such as Claude Haiku 4.5, could still be vulnerable to similar exploits. This suggests ongoing challenges in securing AI systems against jailbreak techniques. The Decoder reported these developments as part of Anthropic’s efforts to strengthen its AI safety measures.
For Japanese investors and tech firms, the resumption of Fable 5’s distribution highlights the increasing importance of robust AI security, especially as AI products continue to enter global markets, including Japan’s growing AI and FX trading sectors.