Britain's privacy watchdog is asking questions about Meta's AI-powered smart glasses after reports that human contractors reviewing recordings from the devices were exposed to extremely private moments captured by unsuspecting users.
Ray Ban Meta smart glasses on sale in a retail display case. Pic credit: Columbo.photog/ Shutterstock
The Information Commissioner's Office (ICO) confirmed it is contacting Meta following an investigation by Swedish outlets Svenska Dagbladet and Göteborgs-Posten that claims outsourced workers tasked with improving Meta's AI systems routinely review footage showing everything from everyday conversations to far more intimate scenes.
The fuss revolves around Meta's Ray-Ban smart glasses, ordinary-looking frames stuffed with cameras, microphones, and an AI assistant that can take photos, shoot video, and respond to voice commands. Meta's terms note that some interactions may be reviewed by humans to improve the system, but according to the Swedish investigation, that review queue can occasionally include moments wearers likely didn't expect strangers to watch.
According to interviews with dozens of workers employed by a Meta subcontractor in Nairobi, Kenya, their job involves labeling and reviewing video, audio, and transcripts collected from the glasses so the company's AI models can better interpret real-world scenes and conversations.
Some of the workers interviewed claim the review queue isn't just harmless AI prompts. Some clips show people getting dressed or using the toilet, while others capture private conversations about relationships, politics, or alleged wrongdoing. Others interviewed by the Swedish outlets claimed the clips occasionally include things like bank cards, personal paperwork, or other identifying details inadvertently caught on camera. As one employee put it: "We see everything."
The investigation raises questions about cross-border data flows. Under the EU's GDPR, companies transferring personal data to contractors outside the bloc must ensure the information is protected through approved safeguards.
This has, unsurprisingly, caught the attention of the UK's ICO. In a statement to the BBC, the watchdog said it was writing to Meta after the claims surfaced, describing the allegations as "concerning." The regulator added that organizations deploying products that capture personal data must be transparent about what information is collected, how it is used, and who may have access to it.
"The claims in this article are concerning," the ICO said. "We will be writing to Meta to request information on how it is meeting its obligations under UK data protection law."
Meta, for its part, told the Beeb that recordings are only used to improve its AI systems in certain circumstances, such as when users choose to share interactions to help train the technology. The company said users can manage their data through device settings and delete recordings at any time.
Neither the ICO or Meta responded to The Register's questions.
The report is yet another reminder that "AI-powered" often still means humans somewhere in the loop – sometimes watching more than users bargained for. ®
Source: The register