Artificial intelligence company Anthropic said it identified an Iran-linked threat actor using its Claude AI system to collect and analyze publicly available information about U.S. naval forces operating in the region. The activity was disclosed in the company’s September 2026 threat intelligence report.

According to Anthropic, the actor used Claude to help build a computer program that tracked naval positions and produced targeting handbooks and recommendations. The collected information reportedly included publicly accessible ship and aircraft transponder identifiers, commercial satellite-imagery searches, U.S. personnel details taken from military photographs, and websites showing naval movements.

The actor also allegedly used the AI system to research known cybersecurity weaknesses in maritime satellite terminals, communications equipment, and industrial control products. Anthropic described the activity as military reconnaissance but did not identify the person, organization, or government agency behind it.

Anthropic said it banned the account, developed new detection measures, and shared threat intelligence with government authorities. The report did not confirm that the compiled information resulted in an attack or that any U.S. Navy vessel or system was compromised.