Anthropic Says Iran Used Its American AI Model to Target U.S. Navy Warships

Live updates

1. 3d ago

   Anthropic Disrupts Iranian Use of Claude AI for US Navy Targeting

   Anthropic has disrupted Iran-linked operators who used its Claude AI model to track US naval forces and compile targeting files. The AI company reports that these actors researched vulnerabilities in shipboard systems and gathered intelligence on Israeli targets. Beyond military applications, the Iranian actors used the model to spread propaganda, spy on dissidents, and surveil Israelis and diaspora Jews. Other Middle Eastern actors in Yemen and the UAE also utilized the model for influence campaigns and war planning.

   Why it matters

   The use of commercial AI for military intelligence highlights the difficulty AI developers face in preventing state actors from bypassing safety guardrails. This activity occurs amid heightened regional tensions involving US naval assets and Israeli security. Anthropic is also monitoring attempts to use its models for biological weapons research.

   What is confirmed

   • Iran-linked operators used Anthropic's Claude AI to track US naval forces and build targeting files.
   • Iranian actors used the Claude model to gather intelligence on Israeli targets and surveil diaspora Jews.
   • Anthropic disrupted Iranian use of its AI for propaganda and spying on dissidents.
   • Military and government actors in Yemen and the UAE used Claude for war planning and influence campaigns.

   Still unconfirmed

   • Iran used Claude to plan the funeral of Khamenei.
   • Yemenis used vibecoding for ballistic-missile guidance and test-fired a guided rocket.
   • Anthropic unveiled a commercial espionage platform profiling social media users in the Gulf.

   What to watch next

   • Evidence of operational deployment of AI-guided munitions by Middle Eastern actors
   • Updates on Anthropic's technical measures to block state-sponsored military intelligence gathering
   • US government response to the compromise of naval targeting data via commercial AI
   Sources used for this update (16)

   1. Haaretz — Live updates • IDF destroys underground Hezbollah complex at Lebanon's Ali al-Taher
   2. WSJ — Anthropic Says Iran Used Its American AI Model to Target U.S. Navy Warships
   3. qz.com — Iran-linked operators used Anthropic's AI to build targeting files on U.S. warships
   4. The Siasat Daily — Iran used Claude to track US warships, plan Khamenei’s funeral
   5. Hana Human Rights Organization — Exposure of Digital Surveillance of Dissidents by the Islamic Republic Using Artificial Intelligence
   6. صحيفة سبق الإلكترونية — AI Monitors the Gulf: Anthropic Unveils a Commercial Espionage Platform Profiling Social Media Users
   7. Livemint — US-Iran war: Anthropic reveals Iranian operators used Claude to gather intel on US naval forces, Israeli targets
   8. Haaretz — Anthropic Says Iran-linked Actor Used Claude to Surveil Israelis, Diaspora Jews
   9. The Jerusalem Post — Tehran used Claude to target US Navy, Israeli individuals
   10. rferl.org — Anthropic Disrupts Iran’s Use Of Claude To Spread Propaganda, Spy On Dissidents
   11. Navy Times — Iran used Claude to target US Navy in Middle East, Anthropic says
   12. Iran International — Anthropic says Iran-linked accounts used Claude for propaganda, US naval targeting
   confidence 90%