AI Policy Wiki
Dashboard
⚲

Alvarez v. Meta Platforms

high confidence · updated 2026-09-13

Proposed class action filed September 4, 2026 in N.D. Ill. alleging Meta extracted biometric information from Facebook and Instagram images without notice or consent, to build NameTag — an unreleased face-recognition system for its smart glasses — and to train and operate text-to-image models including Emu, CM3leon, Chameleon and Muse Image. Pleads five claims across three proposed classes: Illinois BIPA sections 15(a) and 15(b), California right of publicity, California Civil Code section 3344(a), and the California constitutional right to privacy.

Alvarez et al. v. Meta Platforms, Inc., No. 1:26-cv-10773, is a proposed class action filed on September 4, 2026 in the United States District Court for the Northern District of Illinois, Eastern Division, by two parents and their minor children, residents of Illinois and California. It alleges that Meta harvested biometric data from images uploaded to Facebook and Instagram, without the notice or consent required by law, both to build NameTag — an unreleased face-recognition system for the company's smart glasses — and to train and operate its text-to-image generative models, named in the complaint as Emu, CM3leon, Chameleon and Muse Image (Class action complaint, Alvarez v. Meta Platforms (N.D. Ill., No. 1:26-cv-10773, filed September 4, 2026); Source: wired.com).

The complaint runs to 66 pages and pleads five claims across three proposed classes. Three of the five rest on California law, so the suit is wider than the biometric-privacy action the initial coverage described.

Parties (plaintiff)Francisco Alvarez and F.M.A., a minor, by and through his guardian Francisco Alvarez; Jeremy Wahl and I.W., a minor, by and through her guardian Jeremy Wahl
DefendantMeta Platforms, Inc.
Case No.1:26-cv-10773
CourtN.D. Ill., Eastern Division
FiledSeptember 4, 2026
ClaimsIllinois BIPA §§ 15(a), 15(b); California common-law right of publicity; Cal. Civ. Code § 3344(a); California constitutional right to privacy
Class periodSeptember 4, 2021 through the present
Certification sought underFed. R. Civ. P. 23(a), (b)(2), (b)(3)
StatusActive; no ruling recorded

Claims

Five claims are pleaded, each tied to a specific class:

ClaimBasisBrought for
1Illinois Biometric Information Privacy Act § 15(a)Illinois Class
2Illinois Biometric Information Privacy Act § 15(b)Illinois Class
3California common-law right of publicityNational Class; alternatively California Class
4California Civil Code § 3344(a) — commercial misappropriationNational Class; alternatively California Class
5California Constitution — right to privacyCalifornia Class

The prayer for relief at Section VII asks for certification of all three classes under Rule 23(a), (b)(2) and (b)(3); a declaration that Meta's actions violate BIPA §§ 15(a) and (b); "statutory, compensatory, punitive, exemplary, consequential, general, and nominal damages in an amount to be determined at trial" for the National and California Classes; for the Illinois Class the greater of actual damages or $5,000 per intentional or reckless BIPA violation and $1,000 per negligent violation; permanent injunctive relief prohibiting the practices alleged; and fees, expenses and interest. The per-violation figures are BIPA's own statutory amounts, restated in the prayer rather than derived from it. The injunctive relief is pleaded generally in the prayer and specifically in the claims, which ask for an order requiring Meta to publish a BIPA-compliant retention policy under § 15(a) and to comply with § 15(b)'s notice and written-release requirements (Class action complaint, Alvarez v. Meta Platforms (N.D. Ill., No. 1:26-cv-10773, filed September 4, 2026)). A jury is demanded on all issues so triable.

Two alleged uses are pleaded from the same extraction. The first is the training of generative models — Emu, CM3leon, Chameleon and Muse Image. The second is NameTag, an unreleased face-recognition system for Meta's smart glasses, which the complaint treats as identification rather than generation. See AI and Surveillance for the verification-against-identification and targeted-against-mass axes along which such systems are regulated.

The complaint corroborates the NameTag allegations with a Meta patent application published May 21, 2026, U.S. Patent Publication No. 20260142974 A1, "User Identity Verification without Sharing Biometric Data With Platforms," which it describes as disclosing a workflow from smart-glasses capture through face embedding to matching against profile photographs already held by Meta. The pleading acknowledges the patent's on-device, privacy-preserving architecture and argues it nonetheless establishes the capability (Class action complaint, Alvarez v. Meta Platforms (N.D. Ill., No. 1:26-cv-10773, filed September 4, 2026)).

The two biometric theories

The complaint's biometric case operates at two different stages of a model's life, and the second is the novel one.

At the training stage, Section IV.B.3 alleges that Meta's use of Facebook and Instagram images to train and test its text-to-image models "results in the creation and storage of biometric identifiers and information within the models themselves" — that is, that the weights hold the identifier, not merely the training corpus.

At the inference stage, Section IV.B.4 alleges that when a user submits an image of a person's face to one of those models by prompting, the model "likewise creates and stores a biometric identifier of biometric information associated with the person in the image." This theory does not depend on the subject having uploaded anything: it reaches people whose faces were submitted by someone else, which is why each class definition has a third limb covering images uploaded "to any of Meta's generative AI models during the inference stage."

A delayed-discovery, equitable-tolling and fraudulent-concealment section (IV.D) is pleaded alongside, which bears on the limitations defence a five-year class period invites.

Proposed class

Three classes are proposed, each sharing the same three-limb definition — all persons whose image was uploaded to a Facebook account, to an Instagram account, "and/or to any of Meta's generative AI models during the inference stage" — and differing only in geography:

  • Illinois Class (Claims 1 and 2): all such persons in the State of Illinois.
  • National Class (Claims 3 and 4, under California law): all such persons in the United States.
  • California Class (Claim 5, and Claims 3 and 4 in the alternative to the National Class): all such persons in the State of California.

The class period runs from September 4, 2021 through the present — five years before filing. Meta, entities in which it holds a controlling interest, and its legal representatives, officers, directors, assignees and successors are excluded. Certification is sought under Rule 23(a), (b)(2) and (b)(3), with the definitions stated as "subject to confirmation, clarification, or modification based on discovery." The complaint estimates the class could number in the millions nationally (Source: wired.com).

The California claims

Three of the five claims rest on California law and reach past biometric privacy into identity and likeness. The common-law right-of-publicity claim distinguishes storage from extraction, alleging that "[r]ather than merely storing photographs, Meta extracted identity-related information from Plaintiffs' and Class Members' facial images, used those identity attributes to train and improve AI and facial-analysis systems, and incorporated the resulting capabilities into products and services from which Meta derives substantial economic benefit." The § 3344(a) claim recites the statute's remedy of the greater of $750 or actual damages plus attributable profits, and pleads that the use had no connection with news, public affairs, sports or a political campaign — the statutory exemptions. The constitutional-privacy claim rests on a reasonable expectation that a platform "would not transform their photographs and videos into biometric templates, facial embeddings, faceprints, identity vectors, or other machine-readable representations," and on biometric characteristics being "unique, immutable, and capable of identifying a person across platforms, locations, and time" (Class action complaint, Alvarez v. Meta Platforms (N.D. Ill., No. 1:26-cv-10773, filed September 4, 2026)).

Meta's response

A Meta spokesperson said the suit "is without merit and misrepresents our work," that nothing has shipped to consumers, that no final decision has been made on NameTag, and that "we are not building a universal face database" (Source: wired.com).

Context

Meta has twice resolved biometric-data claims at scale: a $650 million settlement in 2020 of an earlier Illinois face-recognition class action, and a $1.4 billion payment to Texas in 2024 over biometric data collection (Source: wired.com).

The suit differs from the training-data copyright cases against Meta in the right asserted. Kadrey v. Meta and Hachette et al. v. Meta (and Mark Zuckerberg) plead copyright in works used as training data; this complaint pleads a statutory biometric-privacy right in the face itself, which does not depend on the authorship of the photograph. The same pairing of a biometric count with training-data allegations appears in the Apple derivative complaint, which adds an Illinois BIPA count over voice recordings alongside its copyright allegations.

Open questions

  • Whether Meta has answered or moved to dismiss, and on what grounds. No response is recorded.
  • Whether the inference-stage theory — that a model creates and stores a biometric identifier when a face is submitted by prompt — has been tested in any prior BIPA action. None is recorded here.
  • How the California right-of-publicity and § 3344(a) claims interact with the biometric claims, given that certification for the California claims is sought on behalf of the National Class rather than the Illinois one.
  • Whether the limitations theory holds. Section IV.D pleads delayed discovery, equitable tolling and fraudulent concealment together, dating discovery to the June 2026 NameTag reporting and the July 2026 Muse Image backlash; a class period reaching to September 4, 2021 depends on it.

Relationships

Sources