Anthropic's annual threat report reveals that the Iranian-backed Houthis used the Claude AI system to develop guided rockets, multi-stage ballistic missiles, and hypersonic vehicles, while the IRGC built an automated identity-profiling harness targeting Israeli officials and Jewish diaspora individuals. The report also details IRGC propaganda operations under the 'Jihad al-Tabyin' mandate, including impersonating an IRGC spokesperson and drafting Supreme Leader funeral logistics.
Anthropic's annual threat report, published this week and highlighted by commentator Emily Schrader, outlines how adversarial actors exploited the company's Claude AI system. According to the report, the Iranian-backed Houthi movement in Yemen used Claude Code to develop software for weapons guidance, navigation, and control, running three parallel development programs: a guided rocket with phone-class flight computer and homing guidance, a multi-stage ballistic missile targeting ranges over 2,000 km, and an R2000 line including a hypersonic glide vehicle.
Separately, the Islamic Revolutionary Guard Corps (IRGC) built an automated OSINT 'identity-profiling harness' that scraped public internet sources to compile intelligence dossiers on hundreds of individuals in Israel and the Jewish diaspora, including Israeli government officials, private citizens, and organizations tied to Jewish communities abroad. The report also describes IRGC 'soft war' propaganda operations under the 'Jihad al-Tabyin' mandate, including generation of doctrine manuals, personas, and target databases. The IRGC impersonated its own spokesperson, fabricated citations to U.S. think tanks during the 2026 US-Israel-Iran war, and drafted logistics for the Supreme Leader's funeral.
The report underscores the growing threat of dual-use AI systems being weaponized by state-backed proxies and intelligence services.
Source and signal
- Open-source intake