Exclusive: Trump tells "The Axios Show" that Anthropic was a national security threat
TL;DR
Trump told Axios he may have viewed Anthropic as a national security threat a week earlier, but not anymore. Axios ties the escalation to an Amazon vulnerability report, Commerce Department export controls and a Pentagon supply-chain risk designation. The administration restricted access to Anthropic's most advanced models from abroad and for foreign nationals inside the U.S. Trump kept Defense Production Act emergency powers on the table, while saying he probably would not need to use them.
Nauti's Take
This is less a pure safety story than a power test between government, cloud partners and frontier labs. Anthropic often positions itself as the cautious player, but caution does not help if Washington thinks risks are being minimized.
The Anthropic statement reads like standard partnership PR; the real test is whether the fight produces measurable jailbreak evaluations and clear access rules.
Briefingshow
A U. S. AI lab is being handled with tools usually aimed at foreign adversaries.
That turns frontier-model safety standards into hard industrial policy: companies that cannot clearly explain jailbreak risk, access controls and their communication with government may face more than reputational damage.
Sources