China has reportedly opened investigations into two of its prominent artificial intelligence companies, DeepSeek and Moonshot AI, following allegations that sensitive user queries were secretly routed to Anthropic’s Claude models.
The investigations, reported by The Information, come after US-based AI company Anthropic accused Chinese AI firms of using proxy networks and fraudulent accounts to access Claude and extract its capabilities at scale. Anthropic said some of the activity involved routing requests from users of Chinese AI services to Claude without their knowledge or consent.
According to Anthropic’s September 2026 report, DeepSeek rerouted more than 12.1 million exchanges to Claude over a 14-day period in July. The company said the traffic included requests from users who believed they were interacting with DeepSeek’s own models.
Also Read | Layoffs Are Rising In Tech, But These Skills Are Helping People Unlock Better Pay
Anthropic also alleged that Moonshot AI, the developer of the Kimi family of models, routed nearly 300,000 customer requests to Claude over a 10-day period through 5,380 fraudulent accounts. Most of those accounts appeared to be located in Singapore and Japan, according to Anthropic.
The allegations go beyond conventional model distillation. Distillation is a widely used AI training technique in which outputs from a larger “teacher” model are used to train a smaller “student” model. Anthropic said the Chinese companies’ activities involved unauthorised access to Claude and the harvesting of its capabilities on an industrial scale.
ALSO READ: At Least One More Fed Hike Likely; RBI Also Set To Raise Rates, Says Nomura — Here’s Why
Anthropic said Moonshot displayed Claude’s responses to users who believed they were using Kimi. The company also alleged that Moonshot captured at least some of those exchanges and used them to extract Claude’s reasoning capabilities for training its own models.
Anthropic further claimed that some of the rerouted requests contained sensitive information belonging to Moonshot customers. The report cited examples involving surveillance data, internal corporate systems and credentials. Anthropic said it could not determine whether Moonshot had informed its customers that their requests were being routed to a third party.
DeepSeek was similarly accused of silently relaying selected user requests to Claude. Anthropic said some requests contained sensitive information from a Chinese technology company, including details related to an AI programme, while other cases involved data connected to a Russian government agency and a Chinese municipal police system.
Anthropic said DeepSeek used a technique involving Claude’s “thinking signature” to obtain reasoning traces that would otherwise have been protected by the model’s safeguards. According to the company, the technique allowed DeepSeek to extract reasoning information for model training.
The broader investigation follows increasing scrutiny in Washington over the practices of Chinese AI companies. Anthropic said it had identified and disrupted distillation campaigns involving seven China-based AI labs since February 2026.
The company also named Alibaba, Xiaomi, Z.ai and MiniMax among firms it said had engaged in various forms of unauthorised distillation. Anthropic said Alibaba alone accounted for more than 151 million exchanges between May and July 2026.
US government agencies have separately raised concerns over Chinese companies’ alleged efforts to obtain US AI model capabilities. Beijing has rejected such accusations, with Chinese officials describing US criticism as an attempt to suppress China’s AI industry.
The latest allegations have also heightened concerns around data privacy. Anthropic said the rerouted exchanges included names, email addresses, corporate information and other sensitive data belonging to users in multiple countries and languages.
The issue comes as Washington and Beijing continue discussions on AI safety and governance. US and Chinese officials agreed recently to continue talks on AI safety and emergency communication protocols, with another meeting expected in Shenzhen within two months.
Anthropic has said it is strengthening safeguards to detect proxy networks, fraudulent accounts and other attempts to extract Claude’s capabilities. It also said it can require identity verification when its systems detect potential abuse, including unauthorised resale of Claude or access from unsupported countries.
Essential Business Intelligence,
Sharp Market Insights,
Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.
