Reports suggest China has copied Anthropic's 'Claude AI' model; Secretary of War Hegseth to meet Anthropic CEO as Pentagon threatens banishment
Anthropic has accused Chinese AI firms DeepSeek, Moonshot, and MiniMax of running industrial-scale "distillation attacks" to steal Claude's capabilities, creating over 24,000 fraudulent accounts and generating more than 16 million exchanges—using proxies to bypass bans on commercial access from China and evade U.S. export controls. The campaigns targeted Claude's core strengths in coding, agentic reasoning, and tool use. Anthropic warns that these illicitly distilled models likely omit safety guardrails, creating major national security risks by enabling potential misuse in bioweapons, cyberattacks, disinformation, or surveillance under Chinese Communist Party influence.
