Meta’s smart glasses spark controversy.

Controversy has once again erupted around a company operating within the AI ecosystem. This time, META is the one in the spotlight.

The controversy is not about how the popular Ray-Ban Meta glasses are used, but about the processes behind them and how the system is trained.

Once again, the issue comes down to transparency and ethics — two of AI’s biggest weak spots.

The "reputational crisis" stems from the fact that META'S content validation centre—responsible for reviewing what users see through the glasses—is based in KENYA, where the process turns out to be far more manual than many might expect. This is not an isolated case, with similar operations taking place across India and Southeast Asia.

Algorithms are still far from perfect, and human input remains essential to training them. In practice, these content validation centres rely on large teams of people who watch, review, approve and reject content.

The problem is that, in KENYA, content reviewers may have access to everything users see through the glasses. And that is extremely difficult to control, because our field of vision—whether we realise it or not—offers a remarkably revealing window into our lives.

The investigation into the "controversy surrounding Meta’s smart glasses" was carried out by two Swedish newspapers, Svenska Dagbladet and Göteborgs-Posten. Their findings reveal that reviewers in KENYA have been exposed to highly sensitive content, including nudity, banking details and intimate moments, ...

At the validation centre, the team helps train the model by labelling objects in individual frames, outlining them, checking descriptions, reviewing transcripts and, above all, assessing whether the assistant has responded correctly to users’ requests.

Of course, the companies operating these validation centres—typically third-party contractors—are bound by strict confidentiality agreements. But, as with the teams responsible for moderating content on social media, that does not change the fact that real people are still viewing the material.

The small print in the Ray-Ban Meta app—the part none of us ever reads—also states that, in order to process and respond to users’ requests, the content captured may be reviewed by automated systems and/or human teams.

Users may have technically accepted this practice through the app’s terms and conditions, making it lawful in principle. However, beyond the ethical concerns, it may also raise questions about compliance with EU data protection law. Personal data belonging to EU users can only be transferred outside the European Economic Area under strict safeguards, so sending this information to Kenya could create significant legal and regulatory challenges.

Irish authorities, where META has its European headquarters, have already launched an investigation and asked the company to provide further explanations.

META is now facing yet another investigation and will have to provide answers—potentially followed by another regulatory fine.

But the deeper issue is not simply that the data may be transferred outside the EU—which could constitute a breach if the required safeguards are not in place. It is the lack of transparency that, once again, surrounds an AI-powered solution and ultimately damages both its positioning and users’ trust in it.