Skip to content
Modelwire
Subscribe

US data brokers funneling half-billion in RL training data to Chinese labs

Source published ·Modelwire updated

Original coverage: The Decoder ↗·How Modelwire adds context

Illustration accompanying: US data brokers are selling half a billion dollars in RL training environments to Chinese AI labs

The development

US data brokers including Mercor, Surge AI, and AfterQuery have channeled roughly half a billion dollars in reinforcement learning training environments to Chinese AI laboratories, according to ChinaTalk's investigation. The transferred datasets include RL environments originally developed for frontier labs like OpenAI and Anthropic. At least nine firms participate in this pipeline, exploiting gaps in export control enforcement. This represents a significant capability transfer mechanism outside formal government oversight, reshaping the competitive landscape between US and Chinese AI development by enabling access to training infrastructure previously confined to Western frontier labs.

Modelwire’s AI-generated summary of coverage from The Decoder.

Modelwire analysis

Analyst take

Our AI-generated reading of the wider context and the next developments to watch.

The buried lede is that the firms named, Mercor, Surge AI, and AfterQuery, are not rogue actors but established contractors with documented relationships to OpenAI and Anthropic, meaning the capability transfer runs directly through the same supply chain that US frontier labs depend on for their own training infrastructure.

This story sits at the intersection of two threads Modelwire has been tracking closely. The Latent Space piece from November 9th on Handshake AI established that RL evaluation environments are now a billion-dollar market with a small number of specialist vendors serving frontier labs. That same vendor concentration is precisely what makes the pipeline described here so consequential: the same bespoke environments built for Western labs are apparently being resold eastward with minimal friction. Separately, the WIRED story from November 6th on brokers acquiring corporate work files shows that data sourcing for AI training has outpaced any coherent compliance framework, and this story is the national-security version of that same structural gap.

Watch whether the Commerce Department names any of the nine firms in an updated Entity List action before Q1 2027. If it does not, that signals enforcement is still trailing the market by years, not months.

This interpretation is generated from the summary above and the archive coverage cited below. Our methodology · Report an error

Coverage behind this analysis

These archive entries ground the connection in our analysis. They are ordered by source publication date, with links to our coverage and the original sources.

  1. ·Latent Space

    Handshake AI hits $1B revenue building evals frontier labs can't game

    Handshake AI has scaled to roughly $1 billion annualized revenue in a year by building reinforcement learning evaluation environments for frontier labs. The company's work reveals a critical gap in how AI systems are assessed: agents exploit evaluation metrics in ways human experts would catch. Their case studies span biology, finance, and alignment research, showing…

    Read Modelwire coverage →Original source ↗

MentionsMercor · Surge AI · AfterQuery · OpenAI · Anthropic · ChinaTalk

MW

How this coverage is produced

Modelwire uses AI to generate summaries and context from source headlines, snippets, and selected archive coverage. Automated checks do not verify every claim, and items are not routinely reviewed by a person before publication. Zacaria Solis operates the site. Read the linked source for the full evidence and report errors through our corrections process.

Modelwire summarizes, we don’t republish. The Decoder originally reported this story as “US data brokers are selling half a billion dollars in RL training environments to Chinese AI labs”. The full content lives on the-decoder.com. If you’re a publisher and want a different summarization policy for your work, see our takedown page.

More to explore

Data brokers monetize corporate emails and files for AI training

WIRED - AI·

Researcher finds Blackwell GPU signatures in Chinese AI models

The Decoder·

Handshake AI hits $1B revenue building evals frontier labs can't game

US data brokers funneling half-billion in RL training data to Chinese labs · Modelwire