The AI- WORLd project is a three-step research project mixing sociological methodology and research, technical assessments and legal analysis. Initiated by a consortium of DiPlab research group members, Reversing.works engineering team and Intérêt à Agir, a french strategic litigation NGO, it focuses on work conditions and personal data of platform workers in order to conduct strategic litigation actions based on GDPR and duty of care legislation.

AI-WORLd is a case study on a work platform hiring "AI trainers" in the European Union. By conducting 30 interviews, collecting over 50 internal training documents, and several worker complaints from California, the first step of the research aims to describe accurately the work conditions and personal data collection (for identification, monitoring, grading and sometimes surveillance purposes) of the "AI trainers" in 12 countries inside the EU.

The second step consists of a technical analysis intended to assess whether there is personal data collection, transfer or processing of any kind by the platform that may infringe GDPR rules.

The final step of the project consists of a series of strategic litigation initiatives throughout the EU to advance platform worker rights. The particular focus on personal data regulation is a means to (1) evidence the abusive data collection practices around the AI industry (2) Engage the responsibility of end-clients using AI services trained in ways that are evidenced to be contrary to EU law and (3) to foster collective action and encourage trade unions to seize the means afforded by GDPR regulation to represent platform workers.

Inspired by the Data Workers Inquiry, the AI-WORLd project also wants two of the "AI trainers" to become co-researchers. Their contributions, from their own research fields, will be featured in the final phase of the report.

Step 1 of the project is in its final step as of April 17 (27/30 interviews, initial report drafted). Step 2 is under way. Step 3 is expected to start in May-June at the latest.

AI-WORLd has received a European AI & Society Fund grant.