Skip to main content
Back to News
Anthropic alleges DeepSeek and Moonshot routed user requests to Claude through intermediaries.
Technology
2 min read
US

Anthropic alleges DeepSeek and Moonshot routed user requests to Claude through intermediaries.

The AMW Read

The allegation meaningfully expands the model-distillation dispute into possible live-query routing and sits within an explicitly described US-China AI conflict.
NoveltySignificance
Foundation Models · Case StudiesGeopolitics
Anthropic
Anthropic

Foundation Models / LLMs

View Company Profile

Anthropic alleges DeepSeek and Moonshot routed user requests to Claude through intermediaries.

Anthropic says DeepSeek and Moonshot covertly sent customer prompts to Claude via a network of intermediary platforms, then presented Claude-generated responses as outputs from their own systems. The allegation appears in Anthropic's 154-page report and, if accurate, would extend a dispute previously centered on model-output distillation into direct handling of live user queries. DeepSeek and Moonshot are described in the report as Chinese AI rivals; the source does not include their response to Anthropic's claim.

The allegation matters because it raises two distinct competitive questions for foundation-model providers: whether a lab can use another provider's outputs to accelerate model improvement, and whether intermediary routing obscures where users' prompts and activity are processed. It also lands amid an intensifying US-China AI dispute: US agencies this week alleged an industrial campaign by six Chinese companies to obtain US model outputs, while Beijing rejected those accusations and threatened retaliation. For Anthropic, the episode reinforces that access controls and customer-data boundaries are becoming part of the frontier-model business model, not merely operational details.

Builders using model gateways or multi-provider orchestration should be able to identify the actual downstream model provider for each request, retain auditable routing records, and obtain clear consent for cross-border processing. Investors should distinguish allegations from established findings, but monitor whether providers respond with tighter platform terms, stronger provenance controls, or restrictions on intermediary access. Per the AI Market Watch index, which tracks roughly 5,000 companies rather than a census, Anthropic has logged 295 pipeline-ingested name matches in the last 90 days, versus 238 in the preceding 90.

#Anthropic #Claude #DeepSeek #Moonshot #FoundationModels

#Anthropic#Claude#DeepSeek#Moonshot#related:Moonshot#related:DeepSeek

How This Connects

Based on Foundation Models · Case Studies

  1. 12h agoAnthropic alleges DeepSeek and Moonshot routed user requests to Claude through intermediaries. · THIS ARTICLE
  2. 4d agoAnthropic assembles $517 billion in compute commitments over 11 monthsAnthropic
  3. 6d agoNvidia is reportedly weighing a $2.5 billion investment in Thinking Machines Lab that would value the startup at roughly $40 billion.Thinking Machines Lab
  4. 1w agoAnthropic Signs Reported $35B Lambda Cloud Deal for Texas AI ComputeAnthropic
  5. 1w agoAnthropic Nears $35 Billion Compute Deal With Nvidia-Backed LambdaAnthropic
  6. 1w agoAnthropic previews MHS for laboratory equipment amid a Pentagon court dispute and a $30T market pitch.Anthropic

Related News

More news from Anthropic

Stay updated with the latest news and announcements from Anthropic.

View all Anthropic news

Discover AI Startups

Explore 5,000+ AI companies with VC-grade analysis, funding data, and investment insights.

Explore Dashboard