Iran‑linked operators used Claude AI to target US naval bases: Anthropic

US-based artificial intelligence company Anthropic said it had identified and disrupted an Iran-linked account that used its Claude AI model to gather and analyse publicly available information on US naval assets and develop potential targeting recommendations.

According to Anthropic, the account used Claude to compile detailed “targeting handbooks” and track US naval positions using open-source information. The activity included the use of a Python pipeline developed with Claude’s assistance to monitor naval movements from publicly available data.

The material collected through the account included the names of US military personnel taken from captions accompanying public military photographs, as well as ship and aircraft transponder identifiers. The account also generated scripts for querying commercial satellite imagery and compiled lists of public websites that provide information on US naval movements.

Anthropic said Claude was further used to examine potential vulnerabilities in shipboard systems, including maritime satellite terminals, Cisco communications equipment and industrial control products. The account also used the AI model to catalogue known Common Vulnerabilities and Exposures (CVEs) affecting maritime VSAT terminals.

The company said it was unclear whether intelligence produced with Claude had been used in any specific Iranian military operation.

The disclosure comes amid heightened tensions between Iran and the United States, following recent Iranian attacks on US naval assets and facilities in the Gulf region. Washington said the attacks caused damage but resulted in no US casualties.

Iranian forces also targeted Naval Support Activity Bahrain, the headquarters of the US Navy’s Fifth Fleet. US officials said the warships evaded the attacks without casualties, although the Bahrain facility was damaged.

Anthropic’s report also detailed a separate account linked to Iran’s Islamic Culture and Communications Organization. The account reportedly uploaded detailed organisational blueprints relating to the funeral of former Iranian Supreme Leader Ayatollah Ali Khamenei, along with succession plans.

Khamenei was buried in Mashhad on 9 July 2026, following a multi-day state funeral after his death in a US-Israeli strike on 28 February 2026.

Anthropic’s findings highlight the growing use of commercially available generative AI tools to process open-source intelligence, automate information gathering and analyse potential military vulnerabilities. The company’s disclosure also raises questions about how AI systems can be misused to accelerate the collection and organisation of sensitive military information that is publicly accessible.

With IANS inputs