Anthropic's repeated AI-risk warnings may have backfired into export restrictions

Anthropic's frequent public statements about the risks posed by advanced AI appear to have contributed to a US decision to restrict foreign access to its newest models, Mythos and Fable, according to analysis by the Financial Times.

The company used risk-related language roughly eight times more often than OpenAI and Sam Altman did in 2026, with five words per 1,000 focusing on risk, regulation, or restrictions compared to OpenAI's 0.6 per 1,000. The FT's research examined official statements, social media posts, and articles from both companies.

The export ban has drawn criticism from prominent AI researchers. Yann LeCun, Meta's former chief AI scientist, said Anthropic's "fear-mongering" about AI had finally yielded results, tweeting: "One reaps what one sows."

The move has alarmed European and Silicon Valley leaders, who worry the Trump administration may use AI-risk concerns as justification for restricting global access to frontier models. It's emerging as a test of how the US intends to manage increasingly powerful AI technologies.

For more context on this developing story, read the full FT analysis.