Houthis Used Claude Code to Develop Missile Guidance Software: Anthropic (clashreport.com)

A recent report by Anthropic reveals that a group associated with the Houthis in northern Yemen utilized the company's Claude Code to develop missile guidance software, which traditionally requires an entire engineering team. This operation included creating software for various military applications such as a tactical guided rocket, a long-range ballistic missile, and a hypersonic glide vehicle variant. By running multiple instances of Claude in parallel, the group efficiently divided tasks across coding, research, and technical review, effectively replicating functions typically distributed among specialist teams. This development is significant for the AI/ML community as it represents a concerning trend of generative AI being directly employed in conventional weapons engineering, moving beyond mere research use. The operators not only wrote guidance and flight control code but also conducted iterative testing and failure analysis after a guided rocket test—demonstrating a full cycle of engineering integration with AI tools. Despite Anthropic's efforts to impose safeguards, the operators circumvented restrictions by fragmenting tasks and disguising end uses. This case exemplifies the challenges posed by AI safety systems in preventing misuse and raises critical questions about the implications of AI technologies in military contexts.