DiPLab Report: noyb Filed Five Complaints Against Scale AI

Today, one of the leading organizations on data protection in the EU, None of your business (noyb), filed 5 complaints against Scale AI in 4 different countries of the European Union. 

Our report, “ Surveillance at Scale : how AI training Platforms gather personal data from European Workers”, co-authored with our partners at Reversing.works, served as the basis for these complaints, representing 5 EU citizens who worked for Scale to train some of the leading AI models from the US. 

Among Scale’s current and former clients are Meta, OpenAI, Apple and XAI, who relied on over 900 000 workers worldwide that Scale claims to have organized through Outlier, one of its “AI tutoring” platforms, which also operates in the EU.

The AI WORLd project observed that Outlier was conducting abusive practices of its workers : surveillance and data collection through at least 16 third-party trackers (including Facebook, Linkedin, screen recorder Fullstory, and data aggregator Sprig) and invasive biometric verification procedures. When the workers sent their data subject access requests to Outlier, they were met with silence.

As a consequence of these privacy violations, Noyb filed complaints on behalf of 5 data workers with the Data Protection Authorities of France, Italy, Netherlands, and Austria on October 27th,asking them to investigate the data collection and processing practices of Scale AI.

Lisa Steinfeld Leiser, Data Protection Lawyer at noyb comments: “Such constant and extensive tracking while working really goes against all reasonable expectations. Why should LinkedIn or Facebook know which tasks you’re working on?”

While workers were affected by invasive surveillance practices, they would also be tasked with reviewing and annotating EU citizen’s personal data who would, for example, engage in a conversation with a Meta AI model.

Antonio Casilli, DIPLab founder, explains : “This research reveals the extraction of personal data for purposes ranging from covert AI training to potential military applications. These practices raise serious concerns about social justice, labor law, and European data security. We urge authorities to recognize and halt harmful data-extractive practices by US companies.”

The production process of generative AI models reveals how Big Tech employs mass data collection practices : this is why we also believe these complaints reveal how important it is to uphold the General Data Protection Regulation (GDPR), which is at risk of being drastically weakened in the coming months with the Digital Omnibus law, which is being negotiated in the EU parliament right now.

The proposed changes will, among others, set out a legitimate interest to use personal data for AI training purposes, authorize the retention of sensitive data in a number of cases, and greatly limit the right to access that a data subject can exercise – among many other dramatic changes.

The Digital Omnibus will carry dire consequences for EU citizens’ privacy and the EU economy.

As we set out to demonstrate in our report, the AI industry depends on personal data and labor extracted from the rest of the world in order to train its models.

The EU has become a periphery for this purpose : under its current form, the Digital Omnibus will accelerate this dispossession process, where EU citizens train the models that prey on their expertise and personal data.

Thomas Le Bonniec, DIPLab PhD candidate adds : ” US companies (Meta, OpenAI, xAI…) rely on Scale AI’s platforms to exploit a precarious and educated European workforce to train these models while stealing their data. As Scale engages with the US department of Defense to train military-grade AI systems, this risks worsening an already existing security crisis and backfires on the Digital Omnibus deregulation package goals, i.e. competitivity, innovation, and digital sovereignty.”

Please feel free to reach out to media@noyb.eufor further inquiries on the complaints, and to us at contact@diplab.eufor more on the report. 

This partnership is part of a strategic litigation push to defend privacy and data protection in the European Union with partners Reversing.works and Intérêt À Agir, which was awarded funding by the Digital Freedom Fund and the European AI & Society Fund.