Meta is confronting a widening legal assault over the personal data that fuels its artificial intelligence ambitions, after a proposed class action alleged the company harvested Facebook and Instagram photos without consent to train AI image-generation models and to build an unreleased face-recognition feature known internally as "NameTag." The complaint lands alongside a separate, sweeping suit from book publishers and bestselling author Scott Turow, who accuse chief executive Mark Zuckerberg of personally orchestrating the use of copyrighted material to train Meta's AI systems.

Together, the cases strike at the core of Meta's strategy: using the vast archive of user-generated content across its platforms — along with scraped books and web data — as free raw material for generative AI. They also revive one of the company's most durable controversies, facial recognition, which Meta publicly retreated from four years ago.

A class action built on photos and faces

According to the proposed class action, Meta illegally harvested photographs uploaded to Facebook and Instagram to train its AI image-generation models and to develop "NameTag," a face-recognition capability designed to identify people in the wearer's field of view. Reporting from WIRED first detailed the existence of the feature, describing it as tailored to Meta's line of Ray-Ban smart glasses, which the company has marketed as a mainstream consumer device since their 2023 launch.

Aggregated coverage of the filings emphasized the scale of the alleged exposure, framing the suit as a violation of the consent of "millions" of bystanders — people who never used Meta's products but whose faces would nonetheless be captured by cameras mounted on other people's heads. That framing is significant: unlike prior biometric privacy battles, which centered on users who had accepted a terms-of-service agreement, this case focuses on non-users who never had any relationship with Meta at all.

WIRED's report and Meta's retreat

In a sequence that has become familiar in Meta's AI rollout, reporting appears to have preceded retrenchment. WIRED subsequently reported that Meta deleted a face-recognition system from its smart glasses app after the outlet raised questions about it. The company has not publicly launched NameTag, and it is unclear whether the code was removed permanently or merely shelved while the feature is reworked.

That deletion could complicate the plaintiffs' case — Meta will likely argue that no biometric identification system was ever deployed to consumers — but it also cuts the other way. The removal of the system after press scrutiny may be cited as evidence that the company understood the feature carried legal risk, or that it was unwilling to defend the practice in public even while building it privately.

Publishers and Scott Turow take aim at Zuckerberg

The second front is copyright. A group of book publishers and Scott Turow, the novelist and longtime president of the Authors Guild, filed suit alleging that Meta reproduced their works at massive scale to train its Llama family of large language models. The complaint goes further than most AI training suits by naming Zuckerberg himself, alleging he "personally authorized and actively encouraged" the conduct.

"Mark Zuckerberg personally authorized and actively encouraged Meta's massive copyright infringement to train AI systems."

Naming a chief executive personally is a deliberate escalation. It mirrors tactics used by authors and news organizations against OpenAI and other AI developers, and it signals that plaintiffs intend to probe internal decision-making rather than treat the infringement as an engineering-side accident. Meta has argued that training AI on publicly available text is transformative and therefore permissible under fair use.

Differing frames, one underlying question

The two matters are being covered through distinct lenses. Technology and privacy outlets, led by WIRED, treat the story as one about consent, biometric surveillance and the physical world — what happens when cameras on glasses meet databases of faces. Publishing-industry coverage, such as Variety's, treats it as a copyright and labor story about authorship and compensation. Aggregator accounts have emphasized the most visceral detail: millions of unwitting bystanders.

Yet both strands converge on the same question — whether Meta may convert other people's data into AI capability without asking, and without paying.

A long history of face-recognition fights

Meta is not new to this arena. In 2020 the company agreed to pay $650 million to settle an Illinois class action under the Biometric Information Privacy Act over its "Tag Suggestions" feature, and in 2024 it settled a biometric privacy suit brought by the state of Texas for $1.4 billion. In November 2021 Meta announced it would shut down the face-recognition system on Facebook and delete more than a billion facial recognition templates, citing concerns about the technology's misuse and the absence of clear regulation.

That history makes the NameTag allegations particularly sensitive. Critics argue it shows the company retreating only when legally forced, then quietly rebuilding the same capabilities on new hardware where the legal framework is less settled.

What happens next

The class action will first have to clear certification — a high bar when the class includes non-users whose faces were never enrolled in any system. Discovery will likely focus on internal documents about NameTag's development, data sourcing for image-generation models, and any risk assessments conducted before the feature was built.

On the copyright side, Meta has seen mixed results in earlier litigation. In the Kadrey v. Meta case, a federal judge found that training Llama on books was fair use but allowed a narrower claim about market dilution to proceed — a split outcome that leaves both sides with arguments.

For Meta, the stakes extend beyond damages. Its smart glasses are a flagship bet on the next computing platform, and face recognition is the feature most likely to make such glasses socially radioactive. Regulation of biometric data remains a patchwork of state laws, with Illinois, Texas and a handful of others leading and no federal standard in place. How these cases resolve could determine whether the camera-equipped glasses on millions of faces become a mainstream product — or a liability the company can never quite put on the market.