Scale AI on Wednesday rejected privacy complaints filed against it in four European countries, calling the claims unfounded and inaccurate and saying its platform’s practices were disclosed to workers and required their consent. The response, given in an emailed statement to Reuters, is the company’s first public answer to complaints filed on Tuesday by privacy group noyb, which alleged that Scale’s Outlier platform illegally monitors more than 900,000 freelancers who produce training data for AI models.
What Scale Said In Response
Scale’s statement was short, but it addressed the central accusations directly. The company said it uses standard tools to verify identities, prevent fraud, protect confidential project information and ensure accurate payments. Those purposes, in Scale’s telling, explain much of the monitoring that noyb described as surveillance. Contributors know their pay rates before accepting tasks, Scale added, and those rates reflect skills, project details and location.
One detail in the response goes to a specific complaint about screen and activity recording. A company spokesperson said certain projects use a separate tool that records time spent and screen activity, which contributors opt into and manually activate before starting work. That framing matters because consent is at the heart of the dispute: noyb argued the monitoring was unlawful, while Scale says the practices were disclosed and required consent.
Scale also denied two further claims that often worry workers and regulators alike. The spokesperson said Scale does not sell workers’ personal data to third parties, and does not use it to train customers’ AI models. Those denials, if accepted by regulators, would narrow the case considerably, because selling personal data or feeding it into customer models would raise a different and more serious set of questions than internal quality and payment checks.
What Noyb Alleged
The complaints, filed in France, Italy, the Netherlands and Austria, target Outlier, the platform where freelancers complete tasks that are recorded to create AI training data for clients including Meta, OpenAI and Google. Noyb asked regulators to fine Scale AI and to examine how the platform assigns tasks. The group alleged that up to 19 trackers followed workers while they worked, recording keystrokes and mouse movements every few seconds, and that an investigation found the resulting file could be rebuilt into a video replaying a work session from start to finish.
The complaints also challenged identity checks. Workers must submit ID documents and face video to identity verification firm Persona, according to the complaints seen by Reuters, which said the checks were invasive, repeated without explanation, and broke European Union rules on biometric data. Noyb further pointed to game-like mechanisms and bonuses in the platform, and said four workers who asked Outlier for copies of their data between December 2025 and May 2026 received no answer.
These are allegations in complaints to regulators, not findings by a court or authority. Regulators in four countries must now decide whether to investigate, and what the evidence shows about disclosure and consent.
Why The Dispute Is Bigger Than One Platform
Outlier sits at a sensitive point in the AI supply chain. Frontier models are trained with large amounts of human-labelled data, much of it done by freelancers across countries rather than employees in one office. How those workers are monitored and paid is therefore a question about the labour behind AI, not just one company’s software settings.
The case also tests how European privacy rules apply to modern, remote, task-based work. Identity verification, fraud prevention and accurate payment are legitimate aims, and platforms argue they need some monitoring to run fairly. Workers and privacy advocates argue that continuous tracking, keystroke recording and repeated biometric checks go far beyond what those aims require, especially when workers depend on the platform for income and may feel unable to refuse. Regulators will have to draw that line, project by project, rather than with a single blanket rule.
Scale’s emphasis on opt-in activation matters here. If the screen-recording tool is genuinely separate, disclosed and manually switched on, it looks very different from background surveillance. If regulators find workers were not told clearly, or that refusing cost them work, Scale’s defence will be much harder to sustain.
Why It Matters
This matters because the people who label, check and improve AI systems are becoming a large, mostly invisible workforce, and the rules governing their treatment are still being written. A ruling against Scale could force changes across data-labelling platforms, from how identity is verified to how much activity can be recorded and how task assignment is explained. A ruling in Scale’s favour would instead give platforms more room to argue that disclosed, consent-based tools for fraud prevention and payment are lawful, even when they are intrusive. Either way, the same questions about consent, control and accountability run through Scale’s own work on AI safety testing, bots flooding a public consultation and warnings about sharing sensitive information with AI systems: powerful systems depend on human data, and the people supplying it want a say in how it is collected.
Background
On Tuesday, noyb filed complaints against Scale AI in France, Italy, the Netherlands and Austria over alleged monitoring of Outlier freelancers. On Wednesday, Scale AI rejected the claims as unfounded and inaccurate, saying its practices were disclosed, required consent, and did not involve selling workers’ personal data or using it to train customers’ models.
Source: Reuters report, earlier Reuters report
0 Comments