AI
Trump's AI Enforcement Gap: How Unwritten Rules Are Reshaping the Industry
The White House forced Anthropic to disable its advanced Claude models over national security concerns, but without clear guidelines—creating a de facto licensing regime that other AI labs now must navigate.
WIRED — AI
The Trump administration's week-long standoff with Anthropic over its Claude Mythos and Fable 5 models has exposed a critical tension in American AI governance: the government is now enforcing rules that don't formally exist.
The saga began when the White House issued an export control directive forcing the AI lab to take its most advanced models offline. Anthropic maintains it violated no explicit procedures, while the administration contends the company acted recklessly. The dispute remains publicly opaque—the government has never officially stated what Anthropic did wrong.
The Real Triggers#
US officials grew concerned when Anthropic shared Mythos with SK Telecom, a South Korean firm the government alleges has ties to China. Amazon's Andy Jassy separately raised concerns that safeguards on Claude Fable 5 could be circumvented. Yet even these potential national security issues were mishandled, according to the reporting: Anthropic says it coordinated the SK Telecom arrangement with the US government beforehand, and it revoked access immediately when concerns were raised. On jailbreaking—the broader security issue—every large language model is vulnerable; expecting a complete fix sets an impossible standard.
The Unwritten Regime#
By forcing Anthropic to ban all foreign nationals from accessing its models, the White House has likely hampered the innovation it claims to champion. Meanwhile, other AI labs—OpenAI, Google, Meta—are drawing a clear lesson: preempt trouble through early access and constant communication with the administration.
This has effectively created the exact "mandatory licensing regime" the tech industry fought against in Trump's recent AI executive order, which officially promised a "voluntary" system. One former White House technology official put it bluntly: "The Trump administration should not have said that this was a voluntary regime. It seems very clear that what they are now doing is a licensing regime."
The result is ad hoc enforcement in real time, without systematic policy or public guidance—the Wild West of AI regulation.
