Meta cut more than 1,000 Kenyan contractors after they reported reviewing intimate Ray-Ban Meta footage. A class-action lawsuit followed, and Kenya opened a regulatory probe. Here's what the scandal exposes about every camera-equipped wearable on your face.
Smart glasses sold a quiet, comfortable lie: that what their cameras capture stays on the device. The Meta Ray-Ban contractor scandal — which broke in February, escalated to a class-action lawsuit in March, and saw Meta terminate its Sama contract — affecting more than 1,000 workers — by late April — has burned the lie down. If you wear a camera on your face, somebody you've never met is plausibly watching the footage to teach the AI behind it.
In February 2026, Swedish publications Svenska Dagbladet and Göteborgs-Posten published a joint investigation that traced Ray-Ban Meta footage from the wearer's face to a labelling pipeline run by Sama, a Kenya-based outsourcing firm. Workers told reporters they had seen videos of people changing clothes, using bathrooms, and in some cases having sex — captured by Ray-Ban Meta users who, the workers believed, did not realise the footage was being uploaded for human review.
The story moved quickly. In early March, plaintiffs Gina Bartone and Mateo Canu filed a class-action against Meta Platforms, alleging the company had failed to disclose that captured video was transmitted off-device to a Kenyan subcontractor. Kenya's Office of the Data Protection Commissioner opened a in April. Late in the month, that Meta had terminated the Sama contract; . Meta's framing was that Sama's workers "didn't meet our standards." Sama's response was that Meta never raised a specific performance issue and that the workers had followed the security and operational protocols Meta provided.
Join the conversation — sign in to leave a comment and engage with other readers.
Loading comments...
Tech privacy stories follow a familiar arc: a company collected more than they said, users get angry, the company tightens controls, the headline fades. The Ray-Ban Meta version is different in three respects.
Continuous, opt-in-by-default capture. Unlike a phone camera you point and shoot, smart glasses are by design hands-free and unobtrusive. The hardware works best when the wearer forgets it's recording. That same property makes consent — yours and the bystanders around you — structurally impossible to verify.
The training-data pipeline is the product. Modern multimodal AI is data-hungry; Meta's pitch for the next generation of glasses depends on continuous learning from real wearer footage. Pulling video to human reviewers isn't a bug, it's the loop. The Sama-to-Meta termination doesn't end the loop; it relocates it.
You are not the only person being recorded. Phone-camera privacy law has had decades to calibrate to the reasonable expectations of bystanders. There is no equivalent calibration for AI glasses worn in changing rooms, waiting rooms, or anywhere a person undressed near a wearer.
We've covered Ray-Ban Meta's appeal as a hardware product before — the glasses are genuinely good, and the integration with Meta's models is the best in the category. The privacy posture changes the buying calculus more than the hardware does. Three concrete recommendations:
Assume any Ray-Ban Meta footage you have ever uploaded was reviewable by a human reviewer at some point. Meta's contract change does not reach back into already-collected training data. If you've worn the glasses indoors with other people present, that data is in the pipeline.
For a broader view of how the industry is treating consent and AI training data this year, our state of cybersecurity in 2026 piece holds up — the supply-chain pattern here is the same as everywhere else in 2026's AI stack: data flows further than the consumer sees, and the visibility isn't catching up.
The lesson is structural, not specific to Meta. Any device that captures continuous video and routes it through a model — phones with always-on cameras, doorbells with AI summaries, fridges with interior cameras — will eventually need a human-review pipeline to keep the AI honest. The Ray-Ban Meta scandal is the first time the consumer side of that pipeline became visible. It will not be the last.
If you are going to wear cameras on your face, wear them like cameras.