A major artificial intelligence company says Iranian groups misused its publicly available technology to gather information on US military personnel and develop targeting capabilities against American forces in the Middle East.


Anthropic detailed the activity in a report published this week covering the period from December 2025 through August 2026.


According to the company, an Iranian "threat actor" used its Claude large language model to collect publicly available information and compile recommendations for targeting US Navy forces operating in the region.


“The threat actor used Claude to compile targeting handbooks…to identify and track naval positions based on open-source information,” Anthropic said in the report.


The company said the user, whom it identified as an “actor”, used Claude to assemble information from a range of public sources. This included lists of US service members identified through military photographs, data from ship and aircraft transponders that broadcast identifying codes, scripts for querying commercial satellite imagery to locate specific geographic areas, and information on public websites that monitor the movements of US Navy vessels.


The Iranian user is also reported to have turned to Claude to study potential weaknesses in US Navy shipboard systems that could potentially be exploited, according to Anthropic.


Anthropic said it eventually identified the activity, banned the account and introduced additional safeguards intended to prevent similar misuse. The company also shared information about the case with relevant government agencies.


The report described several other instances in which Claude was allegedly misused, including by a group of Yemeni actors who used the AI system to develop targeting software for guided missiles.


Tool in wrong hands


Concerns surrounding the development and use of AI have also extended to disagreements within the technology industry itself, including among researchers working at major AI companies.


Jacob Coxon, a researcher associated with Anthropic and OpenAI, resigned this week and publicly questioned the ethical direction of the companies in a lengthy post on X.


The post has since attracted more than 100 million views, in which Coxon argued that many researchers involved in AI development share his concerns and believe there is increasingly little time to ensure that advanced AI systems are developed safely.


“Neither company is acting responsibly,” Coxon wrote. “They are racing straight to self-improving superintelligence and gambling with our lives.”


Anthropic has also faced growing scrutiny over its relationship with the US military. The Pentagon had previously used the company's technology but severed ties with Anthropic after the company refused to accept terms that it said could allow Claude to be used for mass surveillance or in fully autonomous weapons.


By Nazrin Sadigova