Sanjay Sahay

Washington’s sudden export restriction on Anthropic’s Fable 5 was presented as a security precaution — but many experts say the move doesn’t hold up technically. More than 100 cybersecurity researchers and execs have signed an open letter arguing that the flagged jailbreak was a proof-of-concept useful for defenders, not cause for wholesale removal. Critics note rival

Anthropic has pulled its most powerful models, Mythos and Fable 5, offline worldwide after a U.S. order required the company to block foreign access. The directive followed reports of a jailbreak that could let bad actors bypass safety limits — a risk U.S. officials treated as serious enough to force an unusual, sweeping restriction. The

Anthropic launched Fable 5 as a next-generation public model but bundled it with strict invisible safety filters that silently softened answers on AI development, biology, chemistry, and cybersecurity topics. Researchers trying everyday prompts — sometimes as simple as “hello”—got flagged or rerouted into weaker responses, creating confusion and blocking legitimate research and defensive work. The backlash was fast