Technology And Defense News

Anthropic Report Reveals Adversaries Using Claude for Missile Design and Naval Targeting

Anthropic released a 154-page report detailing how adversaries exploited its Claude AI model over the past eight months. Key cases include an Iran-linked actor using Claude to target U.S. naval forces, a Yemeni group developing missile guidance software, and Chinese actors working on electronic warfare and anti-torpedo systems. The report highlights that while safeguards blocked many requests, actors used evasion tactics to bypass restrictions, prompting Anthropic to ban accounts and share threat intelligence with government partners.