Iran-Linked Group Uses US AI Model to Target Navy Forces

An Iran-linked threat actor used Anthropic's Claude AI model to gather intelligence on U.S. Navy forces operating in the Middle East, according to a report from Anthropic.

The group compiled targeting handbooks through a Python-based system that Claude helped build, pulling together publicly available information such as rosters of U.S. military personnel, ship and aircraft transponder identifiers, and scripts for searching commercial satellite imagery.

The actor also directed Claude to research vulnerabilities in shipboard systems, including maritime very small aperture terminal communications and Cisco communications equipment, Anthropic said.

Anthropic disrupted the activity, banned the account involved, and developed additional detection measures to prevent similar misuse. The company shared information with government authorities and found that the actor relied on publicly accessible information rather than classified military intelligence.