(Dreamstime)
An Iran-linked threat actor used Anthropic’s Claude artificial intelligence model to gather intelligence and develop targeting recommendations against U.S. Navy forces operating in the Middle East, according to a new report from the AI company.
Anthropic said the group used Claude to compile targeting handbooks through a Python-based system that the AI model helped build, pulling together publicly available information that could be used to track U.S. naval forces.
The material included rosters of U.S. military personnel taken from captions on publicly available military photographs, ship and aircraft transponder identifiers, scripts for searching commercial satellite imagery and websites that exposed information about U.S. naval movements, Anthropic said.
The Iran-linked actor also directed Claude to research vulnerabilities in shipboard systems, including maritime very small aperture terminal communications, Cisco communications equipment and industrial control products, according to Anthropic’s September threat-intelligence report.
Anthropic said it disrupted the activity, banned the account involved and developed additional detection measures designed to prevent similar misuse, while sharing information with government authorities.
The disclosure comes as U.S. and Iranian forces have been engaged in an escalating confrontation around the Strait of Hormuz, including recent Iranian missile attacks against U.S. Navy ships.
The U.S. military said earlier this month that Iranian forces had attacked American warships, prompting U.S. strikes against three Iranian oil tankers. No American personnel were injured in those attacks, according to U.S. officials.
Anthropic’s findings are part of a broader 154-page report examining how artificial intelligence is increasingly being used for military, intelligence, cyber and other operations.
The company said its investigation covered malicious activity identified and disrupted between December 2025 and August 2026, spanning cyber operations, influence campaigns, surveillance, scams and fraud, biological research, conventional weapons development and attempts to copy AI technology.
In the Iranian case, Anthropic said the actor relied on publicly accessible information rather than classified military intelligence, but used Claude to automate the collection and organization of information that could otherwise require significant human research.
Anthropic separately identified Iranian-linked surveillance operations in which Claude was used to analyze hundreds of thousands of social media posts and identify opposition accounts for monitoring.
The company said it assessed with high confidence that the surveillance units were associated with Iranian paramilitary domestic-security entities and found evidence that the actors were responding to tasking from senior Iranian government officials.
The report also identified other efforts by foreign actors to use Claude for military purposes.
Anthropic said a China-linked actor used the model to develop an anti-torpedo weapons proposal for the People’s Liberation Army Navy, including fire-control software and testing procedures.
The company also said users in northern Yemen, in an operation it assessed as likely involving Iran-backed Houthi militants, attempted to use Claude to develop guidance, navigation and control software for advanced rockets and missiles. A subsequent test launch failed, according to Anthropic’s findings.
Anthropic said the incidents demonstrate that increasingly capable AI systems can provide substantial assistance in areas that historically required highly trained specialists.
The company has responded by strengthening its monitoring and safeguards and said it is sharing information about identified threats with government agencies and other technology companies.
Anthropic also said its own testing found that advanced AI models can perform certain intelligence-targeting and weapons-development tasks that previously required scarce, highly trained human expertise.
© 2026 Newsmax. All rights reserved.