The White House has deemed Anthropic's latest AI model 'Fable 5' a national security threat and imposed export controls. Anthropic shut down the model for all customers, raising concerns that the government's decision could hinder AI competitiveness.
The White House deemed Anthropic's latest AI model 'Fable 5' a national security threat and imposed export controls. Anthropic subsequently shut down Fable 5 and Mythos 5 for all customers.
Anthropic had released 'Claude Mythos Preview' in April with advanced hacking capabilities, later developing safeguards for the public version Fable 5. The White House acted based on a potential jailbreak method discovered by Amazon researchers.
This action reveals inconsistency in the Trump administration's AI policy. By imposing strict regulations under the guise of safety, the government may hinder U.S. competitiveness, as even American companies and the government itself cannot use cutting-edge AI.
The article reports that the White House is intensifying pressure on Anthropic by strengthening AI safety evaluations. In HN comments, criticism focused not on the regulation itself but on the alleged exaggeration of the 'Fable' case cited in the report. One commenter noted that Fable initially refused a security vulnerability review but later complied with a 'code modification' request, suggesting it was more of a simple prompt bypass than a sophisticated jailbreak. Overall, the debate centered on the interpretation of specific cases rather than the effectiveness of regulation.