Anthropic Says Yemen Based Group Attempted to Use Claude AI to Develop Guided Weapons

Anthropic revealed this week that a group based in northern Yemen attempted to use its Claude chatbot to support the development of guided weapons systems. In a report detailing efforts to prevent misuse of its AI models, the company said it identified a cell running multiple weapons development programs, including work related to guided rockets and long-range missile systems.

AI Used to Bypass Engineering Expertise

According to Anthropic, the group relied on Claude to help develop technical software components that would typically require a specialized engineering team. The company said it found no evidence the group successfully created a fully operational weapon, though it noted the actors had carried out at least one field test of a guided rocket, which appeared unsuccessful. Following the failed test, the group reportedly returned to the platform to troubleshoot the issue before the accounts were banned.

Comes Amid Houthi Offensive in Yemen

While Anthropic did not name the group directly, the activity is widely believed to involve Houthi linked actors, who control large parts of northern Yemen, including the capital, Sanaa. The disclosure comes as Iran-backed Houthi forces continue an offensive across the country toward the strategically vital Bab el-Mandeb Strait.