Reading at Advanced level
Meta confronts an expanded privacy complaint centered on the data governance of its AI-enabled Ray-Ban glasses, alleging the company transmitted user and ambient video footage to overseas contractors for annotation and model refinement without sufficient consumer notice or consent frameworks. The original March filing in the Northern District of California was amended on August 31 to include a broader class of plaintiffs—individuals recorded by device users who neither purchased nor authorized the glasses, framing the litigation around the absence of meaningful consent mechanisms for incidental data subjects. Counsel for the plaintiffs contends that Meta's privacy-by-design marketing claims, combined with user-facing data retention controls, mask material omissions regarding transmission of footage to third-party annotation contractors and downstream use in machine learning pipelines.
A February 2026 investigative report by Svenska Dagbladet and Göteborgs-Posten, based on interviews with workers at Sama—a Kenya-based data annotation provider—documented exposure to intimate footage captured through the glasses, including recordings in private settings. The lawsuit frames this as symptomatic of a structural information asymmetry: users lack granular visibility into which data subsets are transmitted for external annotation, under what contractual constraints contractors operate, or how footage is subsequently incorporated into model training regimens. Meta has disclosed general contractor partnerships for data annotation, but the amended complaint argues this disclosure fails to communicate either the scope of ambient capture or the nature of material reviewable by overseas personnel. The case crystallizes the tension between the data requirements of modern foundation models and the juridical status of incidental, non-consenting subjects in hardware-mediated AI development.