The Trump administration’s imposition of export controls against Anthropic should serve as a warning to other technology companies that missteps, and a lack of industrywide guidance on what the government considers national security risks, could result in unexpected sanctions.

U.S. Department of Commerce Secretary Howard Lutnick’s June 12 letter telling Anthropic to disallow access to its Fable 5 and Mythos 5 artificial intelligence models to all foreigners—after the company had already worked with the government to test safeguards prior to the technology’s release—exposes a lack of transparency in the government’s case-by-case enforcement approach, experts say.

[...]

While it’s still unclear exactly what national security grounds the government based its export control directive on, Anthony Rapa, an international trade partner with Blank Rome LLP, said it’s not difficult to imagine those concerns being applied to any company of any size, depending on the capabilities or the risks of their technology. Rapa spoke with Law360 before Lutnick’s letter was made public.

“I tend to think this is more a shot across the bow than something that’s unique to Anthropic,” Rapa said. “For the time being, if we do not have crystal-clear public regulations regarding these questions … or the regulations are trying to catch up to the tech, there may be other instances in which [Commerce’s Bureau of Industry and Security] chooses to make use of this power to the extent they perceive there to be a risk. I think that’s something that very well could end up reverberating throughout the AI ecosystem.”

[...]

As for what other leading-edge technology companies can do to protect themselves, Blank Rome’s Rapa said they should be thinking about conducting an export control assessment pertaining to their technology that involves viewing it holistically and asking whether there’s something in the capabilities of the technology itself that already makes it subject to export control.

“Number two, is the AI capable of generating technical data that is export-controlled or that is risky? Like, can you use the AI—I’m going to make up an example—to generate technology that tells you how to make a nuclear bomb or a weapon of mass destruction, or to carry out a cyberattack? So, thinking about what that tech is capable of generating, if it’s gen AI, is something that’s important for companies to think about,” Rapa said.

Companies should also consider the sufficiency of their guardrails and analyze whether they can be defeated. And if they’re defeated, they should consider what’s the worst thing that could happen, thinking, in particular, about weapons of mass destruction-type applications, large scale cyberattacks and activities that could benefit hostile foreign intelligence services or militaries, Rapa said.

To read the full article, please click here.

“Anthropic Export Controls Stir Fear Of Unforeseen Sanctions,” by Madeline Lyskawa, was published in Law360 on June 18, 2026.