Anthropic accuses three Chinese AI labs of abusing Claude to improve their own models
Engadget
by Jackson ChenFebruary 23, 2026
AI-Generated Deep Dive Summary
Anthropic has accused three Chinese AI companies—DeepSeek, Moonshot, and MiniMax—of conducting large-scale distillation attacks using its Claude chatbot to improve their own models. Distillation in AI refers to less capable models leveraging responses from more advanced ones to enhance their capabilities. However, Anthropic claims these firms engaged in "industrial-scale campaigns," with over 16 million exchanges involving Claude through approximately 24,000 fraudulent accounts. The company asserts that these actions were designed to shortcut the development of more advanced AI models while potentially bypassing safeguards.
Anthropic identified the companies using IP address correlation, metadata requests, and infrastructure indicators, corroborating their findings with industry insiders who reported similar behaviors. This practice raises concerns about the integrity of AI development and competition in the field. Earlier last year, OpenAI also faced similar issues, leading it to ban suspected accounts involved in distillation attacks.
In response, Anthropic has pledged to upgrade its systems to make such
Verticals
techconsumer-tech
Originally published on Engadget on 2/23/2026