Iranian military forces and their Houthi allies in Yemen are using American artificial intelligence to help target US Navy ships and develop advanced weapons, according to a September 2026 threat report by AI company Anthropic.
Although Iran’s spiritual leaders often refer to the US as the “Great Satan” to express their spite, the country’s military, surveillance, propaganda units, and allied Houthis have actively embraced US-built AI tools to target US naval assets and develop weapons, surveillance, and propaganda.
In one of the report’s most remarkable findings, an Iran-linked threat actor used Anthropic’s AI model, Claude, for military reconnaissance and to build targeting recommendations against US naval forces in the Middle East, reports Tom’s Hardware.
To construct these targeting recommendations, the actor combined publicly available ship and aircraft transponder identifiers, commercial satellite imagery, and information on US naval movements.
The perpetrator also extracted the names of US military personnel from captions of publicly available military photos and researched security flaws in shipboard communications equipment, including known vulnerabilities in Cobham Sailor VSAT terminals, Cisco communications equipment, and Schneider Electric EcoStruxure systems.
Anthropic stated that it banned the account, introduced additional detection mechanisms, and shared its findings with government authorities.
In northern Yemen, a cell controlled by the Iran-backed Houthis deployed Claude Code to assist three distinct weapons development programmes.
These included a guided rocket using a phone-class flight computer to assist terminal guidance, a multistage ballistic missile targeting a range of over 2,000 kilometres, and an R2000 missile family variant that included a hypersonic glide vehicle.
The Houthi group deployed multiple instances of Claude instead of a group of software engineers to carry out coding, code review, research, and simulation.
Specifically, the AI was used to write guidance, navigation, and control software; integrate an open-source autopilot with a phone-class flight computer; draft control and position-estimation code; tune parameters; build firmware; and run flight simulations.
Although the Houthis are technically not Iranian, they can share their research and development results with their allies and potentially leverage Iran’s industrial capacity to build their weapons.
Beyond naval targeting and weapons development, Iranian security agencies used Claude to create domestic surveillance tools. One security-linked unit used Claude to analyse 155,216 tweets to profile, identify, and surveil 6,388 opposition individuals in a single year.
Another group used the model as an engineering pipeline to develop domestic tracking tools, including the production-deployed “al-Najm al-thāqib” Firefox extension designed to mass-harvest user identities across major social platforms.
Anthropic has banned 16 Claude accounts associated with Iranian paramilitary and domestic security agencies, although it noted that this does not mean all such accounts have been eliminated.
The report also revealed that Anthropic’s Mythos AI model breached “almost all” classified systems of the US National Security Agency (NSA) within hours during a red-team test. Furthermore, Iran-linked actors and the Houthis are not alone, as China and Russia are also actively using Claude for their military programmes.
