Iran-linked group targets U.S. Navy with AI

An Iran-linked threat actor used Anthropic's Claude AI to compile public information about U.S. naval positions and develop targeting recommendations. The account was banned and the U.S. government was notified.

“They're using it to enhance the way they already operate and the tactics or the techniques that they're already used to doing.”

AI makes military tracking much faster

AI can rapidly aggregate satellite imagery, ship tracking data, and public military photos to build a comprehensive picture of U.S. naval movements. It also helps refine missile and drone targeting.

Adversaries bypass AI access restrictions

Claude is blocked in Iran, but threat actors easily circumvent this using VPNs and foreign phone numbers to create accounts. This makes it difficult for AI companies to prevent misuse.

Iran uses AI for propaganda and disinformation

Iranian state accounts used Claude to generate fake videos, create propaganda, and falsely attribute claims to think tanks like Brookings and RAND. Content was spread across X, TikTok, and Instagram.

U.S. military secrecy is harder to maintain

The U.S. has historically maintained an advantage through superior situational awareness, but AI is making it easier for adversaries to understand American operations. The Pentagon is developing countermeasures.

“It is getting much harder for the U.S. military to cloak or to obscure its activities.”

Pentagon plans AI disguise tactics

One proposed countermeasure involves creating AI personas that mimic realistic life patterns and obscure digital exhaust to confuse adversarial tracking. The Pentagon is working to reestablish operational secrecy.

Tech companies unprepared for adversaries

AI companies made their models widely accessible without anticipating how hostile actors would exploit them. National security experts say this reflects a fundamental mismatch between tech industry practices and military threats.

“What did these companies, what did the national security community think would happen when you make these models as accessible as they are?”