Why the U.S. Government Is Controlling Who Can Use Advanced AI Models

The Trump administration has moved to restore access to Anthropic's Fable 5 AI model after a two-week freeze, Reuters reported. The shutdown had blocked foreign nationals from using Fable 5 and another model called Mythos 5.

In mid-June, the Trump administration ordered Anthropic to prevent people outside the United States from accessing these two models, according to AP News. Anthropic confirmed it was working with the government to restore access in a public statement the same day.

Both Fable 5 and Mythos 5 are Anthropic's most advanced models. For companies with teams spread across different countries — or for international research institutions — a complete ban on using these tools created immediate problems. Thousands of global customers depend on these models for daily work.

The approval process has been uneven. On June 26, the government cleared Mythos 5 for restored access, but left Fable 5 in a separate waiting period, CNBC reported. By June 27, that too appeared close to approval.

How the Ban Works

The government order requires Anthropic to check where users are located and block access from outside the United States. The technical details haven't been publicly explained. Companies typically enforce such restrictions through legal contracts rather than software-based blocking, but doing it through software — actually preventing the connection at the moment of use — is more complex and requires significant engineering work.

It isn't yet clear whether the government specified exactly how Anthropic should enforce this, or whether it left the method up to the company.

Why This Matters Beyond Silicon Valley

The broader shift here is that the U.S. government now treats advanced AI models the way it treats military technology and sensitive weapons designs. This framework comes from rules developed decades ago for hardware and software used in defense, where keeping technology out of adversaries' hands made straightforward sense.

Applying those same rules to AI models creates real confusion. The same model that can help a pharmaceutical company discover new drugs could theoretically be used for purposes the government wants to restrict. There's no clean dividing line.

The fact that Mythos 5 and Fable 5 are being treated differently suggests the government is making case-by-case decisions based on how powerful each model is, not imposing blanket rules. That matters: it means every new advanced model Anthropic releases will likely need individual government approval. For a company moving quickly, that adds time and unpredictability to product launches.

For companies using these AI models, the lesson is practical. Two weeks without access to a flagship tool is the kind of supply-chain problem that global organizations now need to plan for. Businesses with teams in multiple countries, or those that rely heavily on a single AI provider, may need backup plans or redundant tools. A month ago, this felt like a distant theoretical risk. Today, it feels real.

Anthropic has described itself as working cooperatively with the government, consistent with its focus on AI safety and its existing relationships with federal agencies. Whether that partnership helped speed up the Fable 5 decision is unclear from public statements. The timeline may simply reflect how long government review takes.