Fable Flak: Anthropic disables global access to its latest AI model after US ban

Anthropic has roiled entire global sectors—from finance to infor­mation technology—with every successive update to its AI model Claude. When its latest one called Mythos was ready, unprecedent­edly, the company stalled the release stating it was dangerously good at finding software vulnerabilities. They only gave it to select compa­nies involved in digital infrastruc­ture to come up with protections as a pre-emptive measure. The good intentions might have backfired. Anthropic released Mythos finally with guardrails and labelled the model Fable 5. But within three days, it had to pull the model off because the US government banned access to it by any foreign national, includ­ing Anthropic’s employees. This was after Amazon CEO’s warning to the government that his researchers had been able to make Fable expose secu­rity flaws despite the guardrails.

Anthropic has brushed off the charges as minor and something other existing models can also do. There is suspicion that this is the next round in a war brewing between the US government and the company for refusing to allow its AI for surveillance and intelligence purposes. Ironically those who used Fable in that brief window of availability had not found it radical­ly different from earlier models in regular use.