White House Blocks Anthropic Models Over Security Concerns

Anthropic, one of the most vocal advocates for AI safety protocols, has become the first major casualty of the regulatory framework it helped promote. The White House blocked the company's latest models over potential security risks, forcing Anthropic to disable them for all users.

According to Business Insider's Natalie Musumeci, the saga began on June 12 when a US order barred foreign entities from using Anthropic's Fable 5 and Mythos models. Rather than maintain separate access tiers, Anthropic shut down both systems entirely. By June 15, Trump administration officials had reportedly met with the company to resolve the export ban.

Why it matters

This episode exposes a fundamental contradiction in the AI industry: the companies best positioned to warn about AI risks are the same ones racing to build increasingly powerful systems. When safety protocols actually constrain business operations, the tension between caution and competition becomes unavoidable.

Six months of whiplash

Anthropic's recent history reads like a case study in conflicting priorities. In late January, CEO Dario Amodei published a 19,000-word essay warning of the "serious civilizational challenge" posed by AI. Less than a month later, the company weakened its foundational safety commitments, citing competitive pressures and insufficient government regulation.

The pattern continued through spring. In April, Anthropic declared its Mythos model too dangerous for public release due to its ability to identify "high-severity vulnerabilities." By early June, the company was calling for a coordinated slowdown among frontier AI labs to let "societal structures and alignment research" catch up with technological advances.

Then Anthropic released Fable 5 on June 9—essentially Mythos with added safeguards. The next day, Amodei wrote that AI development moves at "lightning pace" while policy moves "very slowly." Three days later, those same policies shut down both models.

The coordination problem

The Mythos incident crystallizes the AI industry's central dilemma. Governments are attempting to regulate technology that even its creators don't fully understand. Both tech executives and policymakers worry that slowing down means falling behind competitors—whether rival companies or rival nations.

As Business Insider notes, everyone may agree that action is needed, but no one wants to move first. The real challenge isn't building safer AI systems. It's determining who decides what "safe enough" means and whether those standards can be enforced without crippling domestic innovation.

Anthropic's February dispute with the Department of Defense, which led the DoD to label the company a supply chain risk, foreshadowed this week's events. The company's resistance to certain government uses of its technology ultimately contributed to the regulatory scrutiny that now constrains its commercial operations.

The details were first reported by Business Insider and Politico.