US Orders Anthropic to Block Foreign Access to Top AI Models
Via The Globe and Mail
The Story
The U.S. government issued an export control directive ordering Anthropic to suspend access to its two most advanced AI models, Fable 5 and Mythos 5, citing national security concerns about a jailbreak method that could bypass safeguards and let the models identify software vulnerabilities.
Anthropic complied and pulled the models for all users, but publicly disputed the rationale, saying it disagreed that the finding of a narrow potential jailbreak should justify recalling a commercial model deployed to hundreds of millions of people.
The company was also asked to have AWS revoke access to both models across all regions, while every other Anthropic model remained available.
Pentagon Chief Information Officer Kirsten Davies defended the move, writing that some things are simply more important than revenue cycles and pre IPO valuation, invoking an America First stance.
The order marks a significant escalation in U.S. AI export controls, which have historically targeted chips and hardware rather than direct access to models, and it lands as Anthropic feuds with the government over its refusal to allow military use of its AI for domestic surveillance and autonomous weapons.
Former White House official Dean Ball noted the directive raises thorny questions about citizenship verification, since users outside the United States, potentially including foreign born company staff, would be blocked from Anthropic's latest models.
Anthropic complied and pulled the models for all users, but publicly disputed the rationale, saying it disagreed that the finding of a narrow potential jailbreak should justify recalling a commercial model deployed to hundreds of millions of people.
The company was also asked to have AWS revoke access to both models across all regions, while every other Anthropic model remained available.
Pentagon Chief Information Officer Kirsten Davies defended the move, writing that some things are simply more important than revenue cycles and pre IPO valuation, invoking an America First stance.
The order marks a significant escalation in U.S. AI export controls, which have historically targeted chips and hardware rather than direct access to models, and it lands as Anthropic feuds with the government over its refusal to allow military use of its AI for domestic surveillance and autonomous weapons.
Former White House official Dean Ball noted the directive raises thorny questions about citizenship verification, since users outside the United States, potentially including foreign born company staff, would be blocked from Anthropic's latest models.
Go Deeper
Read the original reporting at The Globe and Mail.
Read Full Story at The Globe and Mail →