Senior Anthropic technical staff met with Department of Commerce officials in Washington, D.C., on Monday, June 15, 2026 [1].

The meeting aims to resolve a growing dispute between the AI company and the Trump administration regarding national security and the control of advanced technology. This conflict centers on an export-control directive that restricted access to the company's most advanced tools.

Government officials ordered Anthropic to suspend access for foreign nationals to two of its latest AI models: Mythos and Fable [2]. The directive followed concerns over "jailbreaks" — attempts to bypass the safety filters of the AI — which the administration believes could pose a cybersecurity risk if accessed by foreign entities [3].

In response to the government mandate, Anthropic disabled the Fable 5 and Mythos 5 models globally [3]. The company sent senior security researchers to the capital to negotiate the terms of the suspension and potentially find a path toward restoring access [1].

The current standoff highlights the tension between the rapid global deployment of AI and the Trump administration's efforts to tighten export controls on critical technology [4]. While Anthropic seeks to maintain its global user base, the Department of Commerce is prioritizing the prevention of adversarial nations from exploiting vulnerabilities in the models [5].

Technical staff are expected to provide the government with detailed briefings on the specific jailbreak concerns and the measures the company has implemented to mitigate those risks [5]. The outcome of these discussions will determine whether the models remain offline or if a compromise on access protocols can be reached [4].

“Anthropic disabled the Fable 5 and Mythos 5 models globally”

This dispute underscores a shift toward aggressive state intervention in AI deployment. By using export-control directives to halt the availability of specific model versions, the U.S. government is treating frontier AI as a strategic military asset rather than a commercial product. The result is a fragmented global AI landscape where access is determined by citizenship and geopolitical alignment.