Anthropic’s Claude AI has reportedly been exploited for over eight months by Chinese artificial intelligence laboratories, according to a new threat intelligence report detailed by The Decoder. The abuse involved utilizing Claude in developing missile software, coordinating drone swarms, and conducting surveillance operations.
The report highlights that during this period, more than 151 million exchanges with Claude were recorded, illustrating the scale of data interaction and potential misuse. Chinese AI entities, including notable names such as Alibaba’s Qwen, DeepSeek, and Moonshot AI, are implicated in mining Claude’s outputs for training data and operational purposes.
This development raises concerns about the security of AI technologies and their unintended applications in sensitive military and surveillance domains. For Japanese markets, where AI innovation and regulatory frameworks are evolving rapidly, such incidents underscore the importance of stringent controls on AI deployment and cross-border data security.