Meta’s AI Training in EU Crosshairs: 3 Lessons for AI Developers

Europe is opening a new front in the battle over digital privacy, this time targeting the rapid proliferation of smart glasses with built-in cameras. The debate, which has escalated dramatically in Brussels and national capitals, centers on whether wearable recording devices can ever comply with the General Data Protection Regulation’s (GDPR) foundational principles of consent, purpose limitation, and data minimization. The immediate trigger was a report earlier this year by Swedish media revealing that subcontractors for Meta in Kenya were reviewing “deeply private” footage captured by the company’s Ray-Ban Meta smart glasses to help annotate content for training artificial intelligence models. The footage reportedly included recordings of people’s bathroom visits, banking details, and even sexual activity—content gathered without the knowledge, let alone consent, of the individuals filmed. The implications are profound. For years, privacy watchdogs and activists have warned that always-on, outward-facing cameras integrated into everyday eyewear would create a surveillance society where no public or private space is safe from covert recording. Unlike smartphones, which are typically held in a hand or pointed deliberately, smart glasses can record almost imperceptibly, stripping bystanders of the ability to object to being captured. The fact that this data was then offloaded to third-party contractors in another continent for AI training amplifies the sense of violation and adds a cross-border enforcement dimension that EU authorities are ill-prepared to ignore. Key figures have been quick to demand action. Veronika Cifrová Ostrihoňová, a Liberal MEP from the Renew Europe group, told POLITICO that with Meta’s smart glasses reportedly among the “fastest-growing consumer electronics in history,” the moment to intervene is now. She warned the technology could be used to “target women in a way that is not wanted and that is not welcome on the European market.” Her political group has formally asked the European Commission to outline what measures can be taken at the EU level. Meanwhile, Eric Leijonram, director general of Sweden’s data protection authority, described the risks as “high” and stressed the urgent need for a societal consensus on when recording is acceptable. The regulatory response is still taking shape. Options range from guidance clarifying that smart glasses that lack clear, visible recording indicators are illegal, to outright product bans or mandatory design changes such as always-on LED lights. A more aggressive enforcement posture could see data protection authorities order Meta and other manufacturers to halt sales or recall devices already on the market—a massive logistical and reputational hit. The controversy could not come at a more sensitive time. Meta has heavily marketed its second-generation smart glasses as a fashion-forward tech accessory, and the product has helped the company reclaim some consumer cool. Apple, too, is widely reported to be developing its own smart glasses, while Google is exploring a return to the category after the 2013 Google Glass debacle. An EU crackdown would therefore disrupt not just one product line but the strategic roadmaps of multiple tech giants. Complicating matters further, the United States administration has repeatedly complained that EU regulations unfairly target American companies, and the smart glasses dispute could become another flashpoint in transatlantic trade tensions. Beyond the immediate regulatory and trade implications, this controversy is a test case for how the EU intends to govern the intersection of wearable hardware and artificial intelligence. The AI Act, agreed in 2024, imposes strict transparency obligations on AI systems, but the question of whether collecting real-world video data for model training—even when processed offshore—is consistent with GDPR is largely uncharted territory. If Europe sets a precedent that such data collection is illegal absent explicit consent from everyone recorded, it would force a fundamental rethink of how AI models are built, pushing development toward synthetic data, on-device processing, and privacy-enhancing technologies. The smart glasses crackdown is therefore not just about one product category. It is the first major regulatory confrontation over embodied, passive data collection in an era when cameras and microphones are disappearing into the fabric of daily life. The outcome will shape the design of future wearables, the evolution of AI training methodologies, and the balance between innovation and fundamental rights for years to come.