Anthropic Accuses Chinese AI Firms of Data Theft: 16 Million Unauthorized Transactions

Data Extracted via Millions of Fake Accounts As competition in the artificial intelligence sector intensifies, Anthropic has leveled serious allegations against China-based technology giants.
The company announced that DeepSeek, MiniMax Group Inc., and Moonshot used data obtained from Claude models through irregular means to develop their own systems.
According to Anthropic's statement, the companies in question violated terms of service by performing more than 16 million transactions through thousands of fake accounts.
Unfair Advantage via Distillation Method The company stated that Chinese AI laboratories gained an unfair advantage through a method known as "distillation." By training their own models on the outputs of more powerful systems, these firms reportedly accelerated their research and development processes improperly.
This situation is said to be heightening concerns in the U.S.
that Chinese firms are benefiting from the capabilities of American AI laboratories for free.
Growing Sectoral and Political Reactions A similar warning was issued by OpenAI earlier this month.
OpenAI alerted U.S.
lawmakers that DeepSeek was leveraging American technology using distillation techniques.
Several high-ranking officials, including White House AI lead David Sacks, have expressed security concerns regarding the use of this method.
Anthropic emphasized that the threat is not limited to a single company or region, warning that the window for taking action against such violations is closing.