Canada Regains Access to Anthropic's Fable AI After US Export Ban
AI-generated image · Bay Street Wire
Anthropic is redeploying its Fable model globally following a nearly three-week suspension triggered by US government export restrictions.
Canadian users can once again access Anthropic’s Fable AI model after Anthropic and US Commerce Secretary Howard Lutnick announced Tuesday that the US government had lifted export restrictions, BetaKit reports. The ban, which began June 12, also affected the Mythos model.
According to Anthropic, the suspension occurred after researchers at Amazon discovered a method to bypass safeguards in Fable 5 to identify software vulnerabilities. Anthropic stated that while Fable 5 is a restricted version of the more powerful Mythos, testing revealed that other, less capable AI models were also susceptible to the same jailbreak technique. The company clarified that the vulnerability did not reveal any "unique Mythos-level cyber capabilities" and maintained that Mythos has no such unique offensive capabilities.
To prevent future occurrences, Anthropic implemented a new "safety classifier" guardrail for Fable 5. If a user's request is blocked by this classifier, they will be notified and redirected to the Opus 4.8 model.
Anthropic began the global re-rollout of Fable on Wednesday. Regarding the more powerful Mythos 5, the company stated access has been restored for specific US organizations. Anthropic is currently working with the US government to expand the Glasswing program, which provides organizations access to Mythos 5. BetaKit noted that the Canadian government had been granted access to Mythos via the Canadian Centre for Cyber Security just over a week before the export ban took effect; however, it remains unclear if Canadian organizations have regained that access.