The U.S. government’s enforcement letter to Anthropic, which effectively forced the company to take its latest AI models offline just before the weekend, should be a wake-up call for U.S. tech companies, AI labs and otherwise.
We will inform you of the latest information on Dengeki news. On Friday afternoon, the U.S. Department of Commerce sent a letter to Anthropic triggering vague export control directives that prohibit access to Fable 5 and Mythos 5 by non-U.S. persons, including Anthropic employees, citing unspecified national security concerns. Anthropic said it believes the letter is related to the model’s guardrail bypass, but since the letter doesn’t provide specific details, it can’t be sure. The letter has not been made public.
In response, Anthropic has stopped offering both top-of-the-line models to all customers to ensure compliance with the directive. As a result, the US government was able to force tech companies to take their models offline through swift and unilateral action that did not require court approval.
Friday’s intervention by the Trump administration shows that the AI industry is not immune to government intervention. This is also a warning to the broader technology industry. Failure to comply will result in closure of you and your product.
Axios cited sources explaining the tense situation between the two major companies over the weekend, saying that “differences in personality” between Anthropic and the Trump administration, rather than technical issues with the AI products, led to the export directive.
New details about the matter that emerged over the weekend cast further doubt on the government’s already shaky reasoning.
Katie Moussouris, a cybersecurity veteran and researcher who founded Luta Security, said in a blog post that Anthropic recently shared with her a private copy of a paper written by a security researcher describing the alleged circumvention of Fable 5’s guardrails. (The Wall Street Journal reported that the author of the paper was an Amazon security researcher)
Muslih’s blog post explained how the researchers triggered the guardrail bypass, but said the bypass itself “should never have triggered export controls.” The big difference is whether you ask an AI model to “review your code for security issues” or to “fix this code.” Even if the questions are asked a little differently, the end result will be almost the same.
“The conduct described in the paper cannot be meaningfully corrected, and any attempt will only weaken the defense model,” Musli said, calling the export control directive hasty, coercive and misguided.
Since then, Muslih and dozens of other top security researchers and experts have called on the Trump administration to rescind the export control order, calling the move to extract advanced cybersecurity capabilities from U.S. network defenders “dangerous.”
Past administrations have made radical decisions regarding knowledge gaps. For example, the language used by the U.S. government in the 2010s to amend export laws to cover cybersecurity tools that could also be used in cyberattacks was so broad that it almost unintentionally criminalized legitimate security and vulnerability research.
However, the Trump administration’s directive appears to be retaliatory.
Tech Policy Press editor Justin Hendricks said the Trump administration’s move is “likely to make foreign capital wary about the reliability of U.S. AI in critical applications.” The message is that U.S. AI companies cannot be trusted to operate without interference from the U.S. government.
The Trump administration has not confirmed the reason for invoking the export control directive. Did officials panic because they misread the report? Did Amazon CEO Andy Jassy say something to government officials that prompted a response, out of alarm or malice? Was something lost in translation, or was this a way to put pressure on Antropic, which already had a rocky relationship with the administration?The White House may not have realized the far-reaching implications of the letter’s demands, and officials are desperate to undo the damage they caused.
In Hendrix’s words, “This climate is one of clouds of suspicion that government officials are picking favorites based on personal and political factors.” The fallout is that it has set a dangerous precedent for how much control the government intends to exercise over the release of American software.
This time, the government viewed Anthropic as a problem. Tomorrow I might be with someone else.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
