Meta Ray-Ban glasses send private footage to outsourced annotators, raising privacy questions
Mar 3rd 2026
Workers in Kenya report seeing deeply private footage from Meta’s Ray-Ban AI glasses, Meta’s policies allow human review of interactions, and EU privacy experts say transparency and consent around this processing appear inadequate.
- Data annotators in Nairobi employed by subcontractor Sama say they review intimate video and audio from Meta Ray-Ban AI glasses, including bathroom scenes, sex and visible bank cards.
- Meta’s terms of use state that interactions with the glasses may be subject to automated or human review and that content can be used to improve or train AI services.
- The glasses require server processing through Meta infrastructure and cannot run the AI assistant fully offline, contrary to some retail staff claims that no data is shared.
- Retail staff in Sweden gave mixed or inaccurate information about data sharing and user control during sales interactions.
- EU data protection experts say there is a transparency and consent problem under GDPR because users may not realise when cameras record or when material is used for model training.
- Meta referred questions to its privacy policy and confirmed processing under its AI terms but did not provide detailed answers about subcontractor audits or where sensitive footage is reviewed.