Editorial illustration for Meta Faces Lawsuit Over AI Training Data, Facial Recognition

Meta Faces Lawsuit Over AI Training Data, Facial Recognition

By AI Daily Post Edited by Brian Petersen, Editor-in-Chief

September 11, 2026 • 3 min read

A group of parents and children in Illinois and California filed suit against Meta last week in federal court in Chicago, accusing the company of mining their Facebook and Instagram photos without consent. The proposed class action centers on NameTag, an unreleased face-recognition feature tied to Meta's smart glasses, and on generative AI models called Emu and Muse Image. Plaintiffs say Meta pulled biometric data from their images to build these systems, violating privacy statutes in both states.

The lawsuit follows reporting by WIRED in June, which found that code for NameTag had been built into the Meta glasses AI companion app, downloaded more than 50 million times. That code was designed to convert faces caught on camera into biometric signatures and check them against a database of faceprints stored locally on a user's phone, one that Meta had set up to receive updates. WIRED could not pin down where those original faceprints came from. The new complaint attempts to fill that gap, pointing to internal claims and a Meta patent to argue the data traces back to ordinary Facebook and Instagram accounts.

A set of parents and their children in Illinois and California filed a lawsuit last week in federal court in Chicago alleging that Meta illegally used their Facebook and Instagram photos to build NameTag, an unreleased face-recognition system for its smart glasses, and to train generative AI models including Emu and Muse Image.

Meta Sued Over Training Data for Its AI and Face-Recognition Systems - WIRED AI

Why this matters

This case lands right at the spot where model training practices meet biometric privacy law, and that's a spot with real teeth in Illinois. BIPA carries statutory damages per violation, which is why plaintiffs' firms keep filing there instead of jurisdictions with weaker protections. If Meta scraped Facebook and Instagram photos to train Emu, Muse Image, and the unreleased NameTag system without separate consent for biometric use, the "you agreed to our terms of service" defense gets a lot harder to run.

For anyone building multimodal models on user-generated photo data, the lesson isn't subtle: face data and general image data aren't the same legal category, even if they're the same pixels. Consent language written for "improving our products" won't cover biometric extraction in states with dedicated statutes. Expect more suits testing exactly this line as companies keep training on social platform archives. Watch how Meta answers the complaint, and whether it argues the training itself, versus NameTag's deployment, falls outside BIPA's reach.