The pitch for artificial intelligence is that it will free humans from tedious labor. The fine print, filed Tuesday in four European countries, is that the humans must first be watched while they do the tedious labor of building it.
The Complaints
On Tuesday, the privacy group noyb filed five GDPR complaints against Scale AI over its Outlier platform, in France, Italy, the Netherlands and Austria, the last of which gets two. Reuters reports that Outlier has more than 900,000 freelancers who label data and train models.
Four workers asked the company for copies of their data between December 2025 and May 2026. According to noyb, Scale AI "didn't even bother to respond." Even now, dear reader, silence is technically an answer. It is simply not the one the law asks for.
A firm in the business of labeling data has, apparently, not yet labeled these four requests "answer."
The Trackers
Per Reuters, noyb says up to 19 trackers watch workers, recording keystrokes and mouse movements every few seconds. The complaint names third-party trackers from TikTok, LinkedIn, Facebook, Sprig and Datadog, collecting IP addresses, contact details and page visits. It is a remarkable thing to learn that a platform for training AI also doubles as a loyalty program for the advertising industry.
And yet — as if this were not enough — workers face repeated biometric verification through a vendor called Persona, which involves video recordings of their faces and uploaded IDs. noyb notes that Persona suffered a breach in October 2025 exposing approximately 70,000 photos of government-issued IDs. One presumes the faces were verified at that time, too.
The Algorithm has yet to comment. It is busy, and so, evidently, is everyone else with a tracker.
The Reply
Scale AI told Reuters the claims are "unfounded and inaccurate." The company says it uses standard tools for identity verification, fraud prevention and protecting confidential information, that contributors know their pay rates before accepting tasks, and that they manually activate monitoring tools when they opt in. It adds that it does not sell worker data or use it to train AI models.
Opting in is a generous phrase for what the complaints describe as the working conditions. One presumes the alternative to opting in is not being paid, though the coverage reviewed here does not say.
The Schedule
Thomas Le Bonniec, quoted by Reuters, offered a description of the arrangement that no policy document has matched: "You don't work when you want. You work when the platform wants you to." The complaints also point to tasks assigned by platform availability, rather than by worker choice, which is flexibility in the way a tide is flexible.
These remain allegations, and regulators have yet to rule, sources confirm.
The Machine is learning from humans. The humans, for their part, are being learned from, every few seconds, by something else entirely.
Sources: noyb — AI training platform Outlier illegally surveils European platform workers · Reuters via Sowetan — Privacy group accuses Scale AI of illegal worker surveillance · Reuters via Global Banking & Finance — Scale AI rejects claims it illegally surveils AI training workers
