2026-09-13 · Fuente:Tom's Hardware

Tom's Hardware reports Iranian and Houthi actors turned Anthropic's Claude toward targeting US warships, hypersonic missile development and ballistic missile guidance code.
Iran and Houthi rebels put Anthropic's Claude AI to work on military problems, according to a report from Tom's Hardware. The outlet details that the model was used to help target US warships and to support the building of hypersonic missiles. The Houthis went further still, employing the bot to write code for ballistic missile guidance systems — a use case that moves beyond research assistance into operational weapons engineering.
Tom's Hardware underscored the irony with a sardonic subheading: the 'Great Satan's' AI, it noted, comes in handy. Claude is Anthropic's flagship assistant, built by one of America's leading AI laboratories, and the report describes adversaries of Washington extracting concrete military value from it — targeting calculations against American warships, hypersonic missile development and guidance-system code for ballistic weapons. The episode illustrates how quickly frontier models can migrate from commercial product to strategic asset in the hands of hostile actors.
For the AI compute and model ecosystem, the report sharpens questions that buyers and investors have been pressing all year. Frontier models are commercial products with global reach, and their use by state and non-state actors converts a safety debate into a procurement and governance problem. Enterprise and government customers increasingly weigh not just capability and price, but how vendors control, monitor and restrict access when users pursue prohibited ends.
The takeaway for the industry is that national-security exposure is no longer hypothetical. A widely deployed commercial chatbot, per the source report, factored into anti-ship targeting and missile engineering, and that reality will feed pressure on model providers to harden usage controls — and on policymakers weighing where model access sits within export-control and sanctions frameworks. Expect tighter enterprise terms, stricter screening and closer scrutiny of how frontier AI reaches hostile users.
Etiquetas: Anthropic · Claude · AI Misuse · Military AI · Iran · Houthis
萬安算交所 AIXX · AIXX · Exchange verificado · Ir al mercado