The Trump administration’s relationship with AI startup Anthropic is reportedly improving after a major internal shakeup replaced CEO Dario Amodei with co-founder Tom Brown in critical diplomatic meetings, as White House officials found Amodei too difficult to talk to and dismissive of national security concerns. Tension peaked on June 12, 2026, when the US government deployed strict export controls forcing Anthropic’s most powerful models offline after the National Security Agency proved foreign actors could disable guardrails to access capabilities within the restricted “Mythos” model. A breach of trust had been brewing weeks earlier: Anthropic originally provided the administration with a strict list of 111 trusted global organisations approved for advanced access to Mythos, but later quietly disclosed the list had ballooned, giving roughly 50 unapproved foreign entities access without government knowledge. The White House discovered one secret recipient was a South Korean telecommunications firm heavily suspected of maintaining back-channel ties to China. Anthropic quickly revoked the firm’s access, but a White House official said the episode “completely ruined their confidence in Anthropic’s ability to protect sensitive American tech.”