American data-labeling startups that supply major U.S. AI companies and government customers are also marketing training datasets to Chinese laboratories developing competing models, according to the report.
The companies reportedly pursued business at this year’s International Conference on Machine Learning in Seoul, South Korea. Their sales teams arrived prepared to meet buyers across the global AI industry, including potential Chinese customers.
The overlap highlights how widely used AI training-data businesses can operate across geopolitical boundaries. Firms supporting OpenAI, Anthropic and U.S. government work may also be helping provide resources for rival systems built in China.
The development underscores the growing importance of data preparation and labeling in the international AI race, while raising questions about how U.S.-linked suppliers serve customers developing competing technologies.
