Anthropic accuses Chinese companies of misusing cloud AI

Updated: 24-02-2026, 03.56 AM

What is the news?

Artificial Intelligence (AI) company Anthropic has made serious allegations against some Chinese AI companies. The company says that DeepSeek, Moonshot and Minimax misused its cloud chatbot. According to Anthropic, these companies misused cloud systems to improve their AI models. The company has described it as a well-planned campaign at the industrial level and has demanded strict action against such distillation attacks.

distillation attack

What is distillation attack?

In the AI world, distillation means that weaker models train themselves with the answers of stronger models. This process is generally not considered wrong, but Anthropic says that it was misused. According to the company, about 24,000 fake accounts were created and more than 1.6 crore exchanges were made. Due to this, an attempt was made to create an advanced model in a shortcut manner by bypassing its security system.

What was found in the technical investigation?

Anthropic said it has directly linked these activities to specific companies through analysis of IP addresses and other technical information. The company says that it is confident that this work was done deliberately in an organized manner and with solid planning in advance. Other companies in the AI sector have also reported similar suspicious and unusual activities, which has further strengthened these serious allegations.

Legal dispute and further preparations

OpenAI had also made such allegations against some companies and banned suspicious accounts. Anthropic has said that it will upgrade its systems to make them more robust and secure, so that such attacks can be easier to detect. At the same time, the company itself is facing a legal case, in which it has been accused of using illegal copies of songs to train its chatbot, the hearing on which is going on.