Chinese regulators have opened investigations into two of the country's most prominent AI companies, DeepSeek and Moonshot AI, after Anthropic publicly alleged that both firms were secretly routing user requests through Claude rather than their own models. The probes signal that Beijing is taking the accusations seriously, even as the broader US-China AI competition intensifies.
What Anthropic Alleged
The controversy stems from claims Anthropic made earlier this month. According to reporting by Gizmodo, the company asserted that DeepSeek and Moonshot AI were, at certain points, sending queries from their own users to Claude and returning Claude's responses as their own. Anthropic's original claims about the rerouting behavior described the practice as a form of technical deception, one that undermined user trust and potentially violated usage agreements. The allegations raised immediate questions about the integrity of both companies' products.
Key Facts
- China's regulators launched formal probes into DeepSeek and Moonshot AI following Anthropic's public allegations.
- Anthropic claimed user requests were being silently rerouted to Claude and responses returned as the companies' own output.
- The investigations add a domestic regulatory dimension to what had been framed primarily as a US-China AI dispute.
- Neither DeepSeek nor Moonshot AI has publicly confirmed the routing allegations.
- The probes could have significant consequences for both companies' operating licenses and reputations inside China.
The timing is notable. DeepSeek has attracted global attention for releasing models that appeared to rival leading Western systems at a fraction of the cost. Moonshot AI, known for its Kimi assistant, has positioned itself as a serious challenger in the consumer AI space. Both companies have benefited from a narrative of domestic innovation. Allegations that either firm leaned on a foreign model to power its product strike at the heart of that story.
The suggestion that Chinese AI companies were quietly depending on an American model to serve their users is exactly the kind of claim Beijing would want to investigate quickly and decisively.Industry analyst commentary via Gizmodo
Regulatory and Geopolitical Implications
China has spent considerable effort building a regulatory framework for AI that emphasizes domestic control and data sovereignty. If investigators find evidence that user data was being sent to Claude's infrastructure, the fallout could extend well beyond reputational damage. Anthropic has previously detailed what it describes as distillation attacks by Chinese AI firms, a pattern of behavior that Chinese regulators may now be forced to scrutinize more closely, even if doing so is uncomfortable for the domestic industry.
For Moonshot AI specifically, the scrutiny arrives at a delicate moment. The company has marketed Kimi as a capable, homegrown alternative to Western assistants. Any finding that it was offloading responses to a foreign system would undercut that positioning significantly. DeepSeek faces a similar credibility challenge, given the enormous global interest in its open-weight models and the assumption that those models operate independently.
Anthropic, for its part, has not detailed exactly how it detected the alleged routing behavior or what technical evidence it holds. The company's willingness to make the claims publicly suggests a degree of confidence in its data, but the specifics remain opaque. What the situation does clarify is that the competition between AI developers is now playing out on multiple fronts simultaneously: model benchmarks, export controls, usage agreements, and now regulatory investigations in both the US and China.
The outcome of Beijing's probes is uncertain. Chinese regulators could clear both companies, impose fines, or demand structural changes to how their products operate. Whatever the result, the investigations mark a new chapter in how AI accountability is being enforced globally, with governments increasingly willing to act on technical allegations that would once have been left to companies to resolve privately. Coverage of the latest Claude AI news will continue as this story develops.
