Anthropic reveals Yemeni militant group used Claude for missile guidance development

Anthropic's latest threat intelligence report details how a militant group in northern Yemen utilized Claude to develop software for missile guidance, navigation, and control. The group replaced human engineers with AI, splitting projects into multiple sessions to evade safety filters. While Anthropic disrupted these operations and noted that a test launch appeared to fail, the group continued to use the AI to troubleshoot the failure. The report also highlights other major threats, including Russian-linked espionage, automated zero-day vulnerability generation by students in Changsha, and large-scale credential theft by the ShinyHunters group.