The Pulse
Brown Map Traces AI Data Work Across 31 Countries
A Brown senior’s interactive map links data-labeling companies, their corporate customers and worker sites in 31 countries. The project offers a view of outsourced labor behind AI while describing its network as illustrative, not comprehens
AI.info Team · Written with AI · fact-checked ·
Thirteen data-labeling companies, 141 corporate customers and worker sites in 31 countries appear in a new interactive map from Brown University senior Cunyan Ma. The map links firms that supply data work to companies that hire them, making visible some of the human labor behind AI systems. Brown University described the project on October 26, 2026, as a public tool for examining those connections.
Tracing the contractors behind AI training
AI systems often depend on people to label examples, transcribe speech, rate chatbot responses or mark features in medical images. Those labels help train models and assess their outputs. Ma’s map focuses on the firms and locations involved in that work, rather than treating AI development as a process that happens only inside a technology company.
Brown reports that the map draws on existing databases of data-labeling firms and their corporate customers, Ma’s own collection of company delivery centers, and news reports about workers’ conditions. The delivery centers serve as a proxy for where workers are located. The map links those sites to labeling companies and their customers; it does not present a count of the workers themselves.
Large companies across social media, technology, retail and defense contract data-labeling work to specialist firms, Brown reports. Some tasks are routine, such as identifying objects in photographs. Others involve reviewing disturbing material so platforms can label it or train systems to recognize and filter it.
Ma calls attention to difficult, low-paid work
Ma describes the outsourcing of difficult, low-paid work to countries in the Global South as “AI colonialism.” She points to Kenya, Indonesia and Colombia as examples of places where workers may be asked to review harmful text or imagery. The project puts those locations beside the commercial relationships that can otherwise remain difficult for the public to see.
“The problem with a lot of current discourse around AI is that it seeks to ignore that.”
— Suresh Venkatasubramanian, Brown computer science professor, Center for Technological Responsibility, Reimagination and Redesign director, and project adviser
His remark points to a tension in how AI is discussed: the systems may appear automated, but people contribute to the data and judgments used to build and evaluate them.
The map shows connections, not a complete census
Ma describes the current version as illustrative, not comprehensive. Its 13 companies, 141 customers and 31 countries show a portion of the contracting network rather than every firm, customer or worker involved in AI-related data work. The map’s use of delivery centers as a location proxy also means it should not be read as a precise register of individual workers.
That distinction matters for readers trying to understand working conditions. The map draws connections across companies and locations, while the reports it incorporates provide accounts of pay and workplace experiences. It does not, as Brown describes it, establish the terms of every contract or the conditions of every worker represented by a company’s sites.
Ma wants workers to find one another
Ma says the map is intended as a public-education tool, and she hopes later versions could help workers identify colleagues spread across different firms and countries. That matters because dispersed contractors may have little information about people doing similar work elsewhere, making collective action harder.
She also hopes to extend the project to other material parts of AI production, including labor connected to the semiconductor industry. For now, the map’s concrete contribution is a set of links among labeling providers, their corporate customers and delivery-center locations—an organized view of a supply chain that Brown says is often obscured by subcontracting and limited public information.