Anthropic Stalls as White House Export Controls Leave AI in Limbo

The standoff centers on a June 12th executive order banning foreign nationals from accessing the Mythos 5 and Fable 5 models. While the Department of Commerce reportedly vetted Fable 5 prior to release, pressure from Amazon CEO Andy Jassy regarding a potential guardrail vulnerability triggered a rapid, restrictive pivot. Security expert Katie Moussouris, who reviewed the vulnerability, argues the administration’s reaction is disproportionate, noting that the flagged behavior is a standard, essential tool for security professionals seeking to patch code vulnerabilities.

Negotiations have shifted internally at Anthropic, with cofounder Tom Brown and policy chief Sarah Heck now leading talks to replace CEO Dario Amodei. Despite the high-level focus, the uncertainty is compounding. Anthropic relies heavily on the revenue from high-token-cost Mythos models to offset massive infrastructure commitments, including a $15 billion annual deal with SpaceX for data center access. Shareholders Google and Amazon remain sidelined, navigating a delicate balance as the administration signals a broader intent to scrutinize powerful models from OpenAI and other industry leaders.

This regulatory tightening creates a strategic vacuum that global competitors are already exploiting. Industry observers warn that while the White House previously campaigned on a deregulation platform, these sudden, ad-hoc export controls are effectively re-imposing the very oversight they once vowed to dismantle. As OpenAI faces similar pressure to delay future releases, the ongoing friction risks stalling American innovation while the international race for AI dominance accelerates.