The EU bars emotion-reading AI only at work and school

The design

Article 5 of the EU AI Act lists AI practices the regulation prohibits outright, rather than merely regulating. Among them is a ban on placing on the market, putting into service, or using an AI system to infer the emotions of a natural person 'in the areas of workplace and education institutions,' with an exception for systems intended for medical or safety reasons, such as therapeutic use. As the European Commission's own timeline states, this and the Act's other listed prohibitions 'became effective in February 2025,' specifically on 2 February 2025. The prohibition is narrow by design: it names two settings, employment and education, and does not extend to a consumer's own home or personal device.

What the evidence says

The regulation's text is the only authority for what this article covers, and it does not mention companion apps, chatbots, or consumer products at all. A companion product that infers a user's mood from their messages and adjusts its replies accordingly, a design several companion apps describe in their own marketing, operates in a setting, personal use, that this specific prohibition does not reach. That is a meaningful limit on how far this article can be stretched as evidence of EU concern about companion products' emotional design specifically, even though the Act elsewhere addresses AI systems that interact with people, through Article 50's disclosure duty.

What it asks of people

Because the prohibition targets institutions, not individuals, it asks nothing of a person using a companion app for their own purposes. It asks an employer or a school not to deploy emotion-inference tools against people in those settings, a category that could include workplace-monitoring or classroom-engagement software built on similar underlying techniques to a companion chatbot's mood-detection features, without itself regulating the companion product.

Privacy and safeguards

The article's medical and safety exception is undefined beyond the phrase itself; the regulation does not list which conditions or safeguards qualify a system for that carve-out. For companion products, the practical safeguard question, what happens to inferred emotional data once collected, is addressed elsewhere in EU law, primarily through the General Data Protection Regulation's rules on special categories of data, not through this article.

• Does the company operating a companion app disclose whether it infers a user's emotional state, and for what purpose, regardless of whether Article 5 applies to it?
• Would the same emotion-inference technique be prohibited if deployed by a school or employer, even though it is lawful in a personal companion app?
• What other EU rules, beyond this article, govern how a company may store or use data about a user's inferred emotional state?

A ban that applies to schools and workplaces says something about where the EU judged the risk of emotion-reading AI most acute. It says comparatively little, on its own, about the companion products built to read and respond to emotion by design.