Anthropic reports that Chinese military-linked actors utilized its Claude model for defense-related projects. One entity drafted specifications for an anti-torpedo fire-control system, while another developed 16 software modules for electronic warfare and air-defense suppression, targeting sites in Taiwan.
The company also identified surveillance operations targeting Uyghur groups in Syria. Actors used Claude to process data from WhatsApp and Telegram, map social networks, identify recruitment targets, and draft deceptive communications in local dialects.
Anthropic accuses several Chinese firms of industrial-scale model distillation to replicate Claude’s capabilities. Alibaba allegedly generated over 151 million exchanges between May and July 2026 to train its Qwen 3.x models, reaching up to 3 million requests daily.
Other companies, including DeepSeek and Xiaomi, are also accused of similar activities. DeepSeek reportedly generated more than 12.1 million exchanges in 14 days, while Xiaomi accounted for over 400,000, using proxy networks and fraudulent accounts to access the service.
