The White House publically touts partnership with industry, yet behind the scenes actions are restricting frontier AI capabilities for government or international customers.

Anthropic has not publicly addressed Trump’s security threat statement or the specific access proposal.

In mid-June 2026, attention rose as reports said the access proposal was blocked, and Google's DeepMind alum John Jumper joined Anthropic amid a broader AI talent shift.

Export controls on dual-use cyber tech have a long, mixed history, with notable episodes like the Crypto Wars and the global spread of spyware under the Wassenaar framework.

As of publication, Anthropic’s export ban remains in effect, with possible outcomes including a White House lift to preserve U.S. AI competitiveness or broader licensing that could raise costs and curb foreign access.

Mythos, touted as a high-impact cyber-defense tool, had limited access before the ban (roughly 150 vetted users), spurred by concerns about a South Korean partner and possible bypassing of safeguards by Amazon researchers.

Axios noted a June 18 analysis highlighting the administration’s selective deregulation, signaling tension between public pro-industry rhetoric and private enforcement.

Trump reportedly met Anthropic’s Dario Amodei, OpenAI’s Sam Altman, and Google DeepMind’s Demis Hassabis at a G7 AI lunch, underscoring the White House as a convening hub for frontier AI leadership rather than a strict regulator.

Anthropic was ordered to restrict export of Mythos and Fable, halting access for about a week after the directive.

Anthropic, founded in 2021 by Dario and Daniela Amodei, has raised billions and develops Claude, positioning itself as safety-focused.

The case tests whether export controls can curb frontier AI like past efforts to limit encryption and spyware, potentially shaping lab practices and global markets.