Anthropic released its latest threat intelligence report in September, accusing China's "Dark Side of the Moon" of not only using Claude's output to enhance its own models, but also secretly forwarding some customer requests that were originally handled by its own AI to Claude without the user's knowledge, accidentally leaking sensitive information from the People's Liberation Army directly to Anthropic's large model Claude. This incident shocked Zhongnanhai, and the Internet went viral that Yang Zhilin, the founder of Dark Side of the Moon, and 15 senior executives were arrested. However, Dark Side of the Moon urgently clarified on the 12th that the rumors were purely fiction.
Anthropic recently released an AI abuse threat report, revealing various forms of abuse of its model Claude, and directly named seven Chinese AI laboratories for their illegal distillation practices, including Alibaba's Qwen, Moonshot, DeepSeek, Z.ai, Xiaomi, SenseTime, and MiniMax.
Among them, the Dark Side of the Moon incident caused widespread discussion. Anthropic pointed out that during the 10 days from May to July, Dark Side of the Moon transferred nearly 300,000 customer requests to Anthropic, the vast majority of which were handled by Claude Opus. The Dark Side of the Moon used 5,380 fake accounts to forward customer requests that were supposed to be about its own Kimi model to Claude, and finally answered as Kimi's character.
A user who was judged to be possibly related to the People's Liberation Army of China thought that he was using a Chinese AI and submitted the monitoring data of a specific person to the system for analysis, asking to determine whether the monitored subject exhibited abnormal behavior. Anthropic pointed out that the relevant information involves hundreds of monitors in Chengdu, including cameras located near People's Liberation Army facilities, research institutions affiliated with China Electronics Technology Group, and a large Chinese state-owned enterprise. Surveillance information that was originally thought to be handed over to China AI for processing was instead transferred to Anthropic's large model Claude in the United States because Chinese AI companies were suspected of using Claude without authorization.
After the incident broke out, Internet celebrity Xiao Cui mentioned on the YT channel "Xiao Cui Current Affairs and Finance" that this incident was obviously the Chinese military looking for spies. There was a person in the monitored department who was suspected of being a collaborator. It was necessary to analyze whether this person was a traitor or a spy. However, all these surveillance data were accidentally leaked to Anthropic's big model Claude. He bluntly said that Dark Side of the Moon will definitely be held accountable by the Emperor (referring to Xi Jinping).
There are also rumors on the Internet that Dark Side of the Moon founder Yang Zhilin and 15 senior executives were taken away for investigation. Netizens bluntly said that Dark Side of the Moon and DeepSeek took a lot of resources and investment, and finally stole American AI and even passed sensitive information to the United States. This is really serious.
However, Dark Side of the Moon immediately issued a statement on the 12th to clarify, pointing out that the information posted online about the founder and employees of Dark Side of the Moon was purely fictitious and was maliciously spread. It has reported the case to the public security organs as soon as possible and will pursue the legal responsibility of the relevant rumor makers in accordance with the law. It calls on the public not to believe the rumors and not to spread them.
