Anthropic has accused two prominent Chinese artificial intelligence companies, Moonshot AI and DeepSeek, of secretly routing user requests through its Claude models without informing their own customers. The allegations, reported by the South China Morning Post, suggest the firms were presenting Claude-generated responses as output from their own systems, a practice that raises serious questions about transparency, terms of service, and the integrity of AI benchmarks in the competitive US-China technology race.
What Anthropic Is Alleging
According to the report, Anthropic identified patterns of API usage that indicated both Moonshot and DeepSeek were funneling queries from their end users to Claude behind the scenes. Rather than their own models generating the responses, Claude was doing the work, with that fact kept hidden from users. Anthropic has not yet detailed precisely how it detected the behavior, but the company appears confident enough in its findings to go public with the accusation.
Key Facts
- Anthropic alleges Moonshot AI and DeepSeek secretly used Claude to handle user requests.
- Users of those platforms were reportedly unaware their queries were processed by Claude.
- The claims were first reported by the South China Morning Post.
- Both Moonshot and DeepSeek are major players in China's fast-growing AI sector.
- The allegations could implicate violations of Anthropic's API terms of service.
Moonshot AI, the startup behind the Kimi assistant, has previously positioned itself as a direct competitor to Western AI leaders. Coverage of the Kimi model's claims to rival OpenAI and Anthropic made waves earlier this year, which makes the current allegations particularly pointed. If Moonshot was relying on Claude to power user interactions, the competitive framing of its own model becomes harder to sustain.
The practice of reselling or rebranding another company's AI output without disclosure is a violation of trust with users and potentially with the underlying provider's terms of service.Industry analyst commentary cited in South China Morning Post
Broader Implications for the AI Industry
The accusations arrive against a backdrop of intensifying scrutiny of Chinese AI firms in Western markets. DeepSeek drew significant global attention earlier this year when it released models that appeared to match top Western systems at a fraction of the reported cost. If some of that performance was supplemented by calls to Claude's API, it would complicate the narrative around DeepSeek's independent capabilities. It also echoes broader concerns explored in reporting on security vulnerabilities tied to Claude in the context of Chinese research activity.
For Anthropic, the situation is a double-edged development. It confirms that Claude's outputs are considered good enough that competitors allegedly want to serve them under their own branding, which speaks to the quality of Claude's model family. At the same time, unauthorized or undisclosed use undermines Anthropic's business model and complicates its ability to control how its technology is used. The company has been investing heavily in safety and deployment policies, and covert reselling cuts directly against those efforts.
Neither Moonshot nor DeepSeek has issued a public response to the allegations at the time of writing. It remains unclear whether Anthropic has pursued or plans to pursue legal action, or whether it has already suspended the accounts involved. The lack of comment from the accused companies leaves several key questions open, including the scale of the alleged diversion and how long it may have been occurring.
The episode underscores a growing challenge for AI providers: as their models become more capable and widely used, policing API misuse across global markets becomes increasingly complex. Detecting covert rerouting requires technical forensics that most companies are still developing, and enforcement across international jurisdictions adds another layer of difficulty. How Anthropic chooses to respond in the coming days will likely set a precedent for how the industry handles similar disputes going forward.
