Anthropic publicly accused three AI companies of carrying out a large-scale distillation attack against its Claude chatbot. In a statement posted on its website, the company claimed that DeepSeek, Moonshot and MiniMax participated in a campaign to extract Claude’s capabilities to improve their own models.
Over 16 million interactions
In the AI sector, distillation refers to the process by which a less capable model learns from the output of a more advanced system. Anthropic said that while distillation can be used legally, it can also be abused in ways that undermine safeguards and intellectual property.
According to the company, three Chinese companies were responsible for more than 16 million interactions with Claude through approximately 24,000 fraudulent accounts. Anthropic claimed that these activities were intended to use Claude as a shortcut to develop more advanced AI systems.
Anthropic said it linked the campaign to a specific company with a high degree of confidence. The company cited IP address correlation, metadata requests and infrastructure metrics as part of its investigation. He also said he collaborated with other AI industry players who observed similar behavioral patterns.
OpenAI has faced similar criticism
Early last year, OpenAI expressed similar concerns, accusing competitors of refining its models and banning accounts suspected of misuse.
Anthropic said it plans to enhance the system to make distillation attacks more difficult to execute and easier to detect.
Anthropic is also facing legal scrutiny while raising concerns about the misuse of its technology. Music publishers filed suit, claiming Anthropic used unauthorized copies of the song to train its Claude chatbot.
Source Link : https://propakistani.pk/2026/02/24/deepseek-and-moonshoot-accused-of-stealing-from-anthropics-claude-chatbot/