Fact-check: “Anthropic Says Iran Used Its American AI Model to Target U.S. Navy Warships.”
Verdict: Misleading (72% confidence)
Anthropic's own September 2026 threat intelligence report confirms that an 'Iran-nexus' threat actor used Claude to compile targeting handbooks, track naval positions using open-source data, and research vulnerabilities in shipboard systems — so the core activity is documented. But the claim overstates on two key points: Anthropic used the label 'Iran-nexus,' not 'Iran' as a confirmed state actor, and the operation involved intelligence-gathering that was detected and disrupted before any attac…
This fact-check was conducted by SpinkillerAI, a non-partisan AI-powered accountability platform.
Related fact-checks
- Divatriculitus is a recurring problem once you have it — Mostly True
- "Pink noise" may lead to more restorative sleep. — Partly True
- Trump administration blocks $25 million in funding for national parks. — Unverifiable
- Trump's $5000 dividend promise would not use taxpayer dollars. — Misleading
- Elon Musk is Running a Fake Voter Registration Site Designed to Steal and Sell Data — Misleading