The US government’s Anthropic models ban was never about an AI jailbreak
The U.S. Commerce Department forced Anthropic to take its Fable 5 and Mythos 5 AI models offline Friday afternoon by invoking an export control directive. The action, cited as a national security concern, banned non-Americans from accessing the models, prompting Anthropic to shut down the services for all customers to ensure compliance.
Why did the U.S. government shut down Anthropic’s AI models?
The U.S. Commerce Department sent a letter to Anthropic on Friday afternoon invoking an obscure export control directive. This order banned non-Americans, including Anthropic’s own employees, from accessing the Fable 5 and Mythos 5 models due to unspecified national security concerns.
Anthropic responded by pulling both top models offline for all customers. The company stated it believes the directive relates to a bypass of the models’ guardrails, though the government has not provided specific details and the letter remains private.
According to Axios, the directive may have been driven by “personality differences” between the Trump administration and Anthropic rather than technical failures of the AI products.
What are the technical disputes over the export control?
Katie Moussouris, founder of Luta Security, stated in a blog post that the government’s action was “hasty, heavy-handed, and misguided.” Moussouris reviewed a private paper describing a guardrail bypass in Fable 5, which The Wall Street Journal reports was written by Amazon security researchers.
Moussouris argued that the bypass—which involves the difference between asking a model to “review code for security issues” versus “fix this code”—should not have triggered an export control. She noted that the behavior cannot be meaningfully fixed and attempting to do so could weaken the model’s defensive capabilities.
Moussouris and other security experts have called on the Trump administration to revoke the order. They described the removal of advanced cybersecurity capabilities from U.S. network defenders as “dangerous.”
How does this affect the broader AI industry?
Justin Hendrix, editor of Tech Policy Press, stated the move is likely to raise alarms in foreign capitals regarding the reliability of American AI for critical applications. He suggested the action signals that U.S. AI companies may not be able to operate without government interference.
Hendrix described a “cloud of suspicion” where senior officials may be picking favorites based on political and personal factors. This creates a precedent for how the government might wield control over the release of American-made software.
What may happen next?
The Trump administration has not confirmed the exact reasoning for the directive. Possible scenarios include a misinterpretation of the Amazon researchers’ report or pressure resulting from a fractious relationship between the White House and Anthropic.
The administration could face continued pressure from security researchers to revoke the order. Additionally, the government may be scrambling to undo the consequences of the letter if officials were unaware of the far-reaching impact of the demand.
Frequently Asked Questions
Which AI models were pulled offline?
Anthropic shut down its Fable 5 and Mythos 5 models.
Who reported the guardrail bypass?
The Wall Street Journal reports that the paper describing the bypass was written by security researchers at Amazon.
Did the government require court approval for this action?
The source indicates the U.S. government forced the models offline through a swift, unilateral action that did not appear to require court approval.
Do you believe government export controls are a necessary security measure or an overreach into private tech operations?