The UK’s data watchdog is seeking answers from Meta after a report claimed outsourced workers were able to view highly sensitive videos recorded by the company’s AI-powered smart glasses, including footage of people using the toilet or having sex.
The Information Commissioner’s Office said the allegations were “concerning” and confirmed it would write to Meta to request information on how the company is meeting its obligations under UK data protection law.
The claims come from an investigation by Swedish newspapers Svenska Dagbladet and Goteborgs-Posten, which reported that subcontracted workers in Kenya reviewed images and videos captured by Meta’s Ray-Ban smart glasses to help improve the performance of its artificial intelligence systems.
Meta said that contractors may sometimes review user content, including films and images, to enhance the user experience, as outlined in its privacy policies. The company said data is filtered to protect privacy, which can include blurring faces. However, workers quoted in the investigation said these protections sometimes failed, allowing them to see people’s faces and highly personal moments.
One worker told the newspapers: “We see everything, from living rooms to naked bodies.”
Users must activate recording manually or by voice command, and the glasses display a light when recording is in progress. However, critics argue many users may not realise that some of this content can later be reviewed by human contractors. Meta’s UK AI terms of service state that interactions with its AI systems may be reviewed either automatically or manually by humans.
The ICO said devices that process personal data, including smart glasses, must give users control and clear transparency. “Service providers must clearly explain what data is collected and how it is used,” the regulator said, adding that it will question Meta directly about the report’s findings.
The workers involved were data annotators employed by Nairobi-based outsourcing firm Sama, whose role is to label and review content so Meta’s AI can better interpret images and conversations. They also reviewed transcripts of interactions with the AI to check the quality of its responses.
The workers described strict workplace rules, including surveillance cameras and bans on mobile phones, but said they were exposed to disturbing and intimate material. One example involved glasses left recording in a bedroom, capturing a woman undressing without her apparent awareness.
Meta said it takes user privacy seriously and warns customers not to misuse the technology, advising them to make others aware when recording and to avoid filming in private spaces.
Concerns about misuse of AI-powered wearable devices have grown as the technology spreads. Women have previously reported being filmed without consent using smart glasses, and privacy campaigners warn that rapid advances in AI are outpacing public understanding and regulation.
The investigation adds pressure on Meta and its partners to demonstrate stronger safeguards for personal data as smart glasses and other AI wearables become more common.