Seventh Circuit Vacates Clearview AI’s Landmark Biometric Privacy Settlement Over Inadequate Class Representation

On July 13, 2026, the U.S. Court of Appeals for the Seventh Circuit vacated a district court’s approval of a novel class-action settlement in In re Clearview AI, Inc. Consumer Privacy Litigation, one of the largest and most closely watched biometric privacy cases in the country.¹ The settlement, which offered class members a slice of Clearview’s future equity in lieu of a cash payment, had already survived one round of objections and won final approval below. But in Weissman v. Clearview AI, Inc., a unanimous panel held that the deal could not stand because no representative negotiated on behalf of the “Nationwide Class” alone, even though that class received far less than several state-specific subclasses under the settlement’s allocation formula. The decision does not fault the settlement’s unusual structure or its reliance on an equity stake rather than cash. It instead turns on a more fundamental point of class-action procedure: adequacy of representation under Rule 23(a)(4), and the requirement that conflicting subgroups within a class each have their own advocate at the negotiating table. For companies and counsel managing sprawling privacy multidistrict litigation with overlapping nationwide and state-law claims, the ruling is a pointed reminder that settlement architecture, not just settlement value, will draw close appellate scrutiny.

The Clearview AI Litigation and Its Equity-Stake Settlement

Clearview AI operates what the Seventh Circuit called “a search engine for faces.” The company scrapes photographs from public websites, including social media accounts, and uses artificial intelligence to generate “facial vectors” that can match an unidentified photograph against its database and return links to every other image of that person online. After a January 2020 New York Times exposé revealed the scope of the practice, plaintiffs filed a wave of putative class actions, which the Judicial Panel on Multidistrict Litigation consolidated in the Northern District of Illinois. The consolidated complaint asserted claims on behalf of a Nationwide Class (declaratory judgment and unjust enrichment) and four state-specific subclasses: an Illinois Subclass invoking the Illinois Biometric Information Privacy Act (BIPA), and California, New York, and Virginia Subclasses invoking each state’s own privacy, misappropriation, or consumer-protection statutes.

After more than five hundred docket entries and two rounds of mediation before a retired federal judge, the parties agreed to a settlement built around a 23% equity stake in Clearview, to be paid out upon an IPO or “liquidation event,” with a settlement master empowered to instead sell the stake or demand a share of Clearview’s revenue if no such event occurred by a set deadline. Notably, none of the eight original class representatives agreed to the settlement, and Lead Class Counsel substituted four new representatives, all of whom belonged to one of the favored state subclasses.² The opinion does not say why the original representatives refused. Lead Class Counsel strongly implied to the district court that the refusals reflected strategic disagreements with the representatives’ own attorneys, who had “variously proposed settling for injunctive relief only (allocating the available cash to attorney’s fees), bankrupting Clearview by continuing to litigate … and/or pursuing the defendant class of Clearview customers.” The court also noted that by that point none of the eight was still represented by Lead Class Counsel, because an attorney had left the firm and taken those clients with him. Under the settlement’s allocation formula, each Illinois Subclass member received ten shares of the settlement fund, each California, New York, or Virginia Subclass member received five shares, and anyone who belonged only to the Nationwide Class received a single share. The district court granted final approval on March 20, 2025, over sixteen objections and opposition from a coalition of state attorneys general, finding the deal fair, reasonable, and adequate under Rule 23(e)(2).³ Two Nationwide Class members, Robert Weissman and Rick Claypool, appealed.

No Inherent Substantive Problem, Reviewed for Abuse of Discretion

Before reaching the representation issue that decided the case, the panel, in an opinion by Judge Hamilton, rejected two substantive challenges to the settlement, finding “no inherent substantive problems” with either feature. That framing matters: the standard of review was abuse of discretion, and the question was whether the district court could approve the deal, not whether the Seventh Circuit would have. The objectors first argued that a settlement lacking any injunctive relief could not be fair, reasonable, and adequate. The court disagreed, explaining that BIPA could not support nationwide injunctive relief because only the Illinois Subclass asserted BIPA claims, that a separate settlement with the American Civil Liberties Union had already enjoined Clearview from disclosing biometric data to private parties outside of BIPA compliance, and that the Nationwide Class’s unjust enrichment theory was too speculative, and too variable across state law, to require an injunction as a condition of settlement.

The objectors also challenged the equity-stake structure itself, arguing that tying class recovery to Clearview’s future success was “fundamentally unfair” because it would make victims into shareholders in the company that harmed them. The Seventh Circuit again disagreed, noting that equity-based settlements carry inherent uncertainty but are not disqualifying, particularly where the alternative, given Clearview’s finances, was a substantial risk that the company would become insolvent before paying any judgment at all. The court found the district court’s estimated $51.75 million valuation of the 23% stake to be “commensurate with other BIPA settlements identified by the court, taking account of differences in defendants’ resources and the size of the classes, among other factors.” That qualifier does the work. The comparison set included the $650 million Facebook biometric privacy settlement,⁴ along with smaller resolutions involving Google, TikTok, and Snap; the court was not equating $51.75 million with $650 million, but measuring the figure against defendants of very different size and solvency. The panel also flagged “one significant concern” for remand: the district court never made a finding as to the anticipated value of the settlement’s cash-demand fallback option, because it never examined Clearview’s revenue statements the way the mediator had. If the parties reach a substantially similar agreement on remand, the court said, the district court must be informed directly about Clearview’s finances and make appropriate findings on that option.

The Fatal Flaw: No Separate Voice for the Nationwide Class

The settlement’s undoing was procedural. The Seventh Circuit held that class-action settlements require “structural assurance of fair and adequate representation for the diverse groups and individuals affected,” a principle the Supreme Court articulated in Amchem Products, Inc. v. Windsor when it affirmed the vacatur of a global asbestos settlement that failed to address conflicts between currently injured claimants and those merely exposed to future risk.⁵ The Seventh Circuit’s own precedent, Eubank v. Pella Corp., similarly treats the requirement that class representatives approve a settlement, as fiduciaries for absent class members, as a core structural safeguard, not a formality.

Here, every class representative who endorsed the settlement belonged to both the Nationwide Class and one of the favored state subclasses, and no attorney was tasked with representing the Nationwide Class’s interests alone. The court found this fatal because the case presented exactly the kind of “zero-sum” allocation dispute in which one subgroup’s larger share necessarily comes at another subgroup’s expense. Citing the First Circuit’s recent decision in Murray v. Grocery Delivery E-Services USA Inc., the panel explained that a common settlement fund like this one “can be a warning sign” for intra-class conflicts, because in zero-sum circumstances a larger portion for one group comes at the expense of the rest.⁶ Murray carries its own carve-out, which the panel quoted: assigning appropriate relative weight to claims with significantly different elements or defenses should happen through negotiations between counseled representatives of the different groups, rather than by the district court, “unless the appropriate relative weight is clear-cut.” The court also relied on Ortiz v. Fibreboard Corp., where the Supreme Court held that a settlement needed to distinguish between claimants with insurance-covered injuries and those without, and on the Second Circuit’s decision in In re Literary Works in Electronic Databases Copyright Litigation, which vacated a settlement dividing copyright claimants into three tiers because no representative held only the lowest-value claims.⁷

The panel rejected the argument that overlapping class membership solved the problem, since every named representative also stood to gain from the same allocation being challenged. It likewise rejected the district court’s reasoning that the allocation ratios “reflected the strength and viability” of the different claims, observing that this may well be true as a matter of relative claim value, but that the point of Rule 23(a)(4) is precisely to ensure an independent advocate tests and negotiates that judgment on behalf of the group receiving less, rather than leaving the court to guess whether the ratio was fair.⁸ Finally, while expressly declining to decide how heavily the point should weigh, the court agreed with the objectors that the wholesale substitution of all eight original class representatives after each refused to agree to the settlement was a potential “red flag” — a phrase the objectors drew from Eubank. The district court noted the replacement but did not otherwise discuss it in evaluating adequacy of representation. On remand, the panel instructed, the district court should consider the substitution and the reasons for it in deciding whether the class representatives adequately protected the class’s interests, as opposed to counsel’s interest in settling the litigation.

A Structural Rule, Not a One-Off

The Seventh Circuit was careful to frame its holding as a general remand rather than a rejection of any particular settlement value or structure; a revised settlement need not abandon the equity-stake model or restore injunctive relief. But the court’s reasoning has implications well beyond this case. Privacy multidistrict litigation increasingly combines a broad nationwide theory, often unjust enrichment or a declaratory judgment claim of uncertain strength, with state law claims of widely varying value, from BIPA’s substantial liquidated damages⁹ to thinner common-law and statutory theories elsewhere. That structure is common precisely because federal courts have grown skeptical of certifying single nationwide classes for claims governed by fifty different bodies of state law. The Seventh Circuit’s opinion signals that once litigants divide a class that way for settlement purposes, they cannot then negotiate the relative allocation among those groups without a representative for each side of the divide, at least where the difference in treatment is not “clear-cut” on its face.

The court did leave itself some running room to avoid endless subclass fragmentation, reiterating that only “fundamental” conflicts, not “hypothetical” ones, require separate representation, and that there must be “an end to reclassification with separate counsel” at some point. It also clarified that the Nationwide Class itself does not need further division by state, since its members share a common interest in maximizing their collective share relative to the subclasses even though their individual claims might, in theory, be litigated under different states’ unjust enrichment law.

Practical Implications for Multidistrict and Class Settlement Design

For companies facing privacy class litigation with both nationwide and state-specific claims, and for the plaintiffs’ firms negotiating against them, the Clearview decision offers several concrete lessons for structuring a settlement that will survive appellate review.

First, counsel should identify, at the outset of settlement negotiations rather than after a deal is struck, whether the proposed class or classes contain identifiable subgroups whose recoveries will differ meaningfully, whether because of different statutory damages, different causes of action, or a common settlement fund that must be divided among them. Where that is likely, as it will be whenever a nationwide theory is paired with one or more strong state statutory claims like BIPA, the parties should build in separate representation for each affected subgroup before the allocation is negotiated, not merely before the settlement is presented to the court for approval.

Second, defendants should be cautious about relying on a single set of class representatives or a single negotiating team to bind subgroups with materially different claims, even where individual class members happen to belong to more than one subgroup. The Seventh Circuit made clear that dual membership does not cure the conflict, since a representative who benefits from a favorable allocation cannot simultaneously and credibly advocate for the group receiving less. Defendants have an interest in a durable settlement, and a structure that invites a later adequacy challenge undermines that interest just as much as it does the class’s.

Third, when a settlement substitutes some or all of the original class representatives, counsel should expect that substitution to draw scrutiny and should be prepared to document the reasons for it. The Seventh Circuit agreed with the objectors that the wholesale substitution of Clearview’s eight original representatives, each of whom had declined to agree to the settlement, was a potential “red flag,” though it expressly declined to decide how much weight the point deserved because the representation defect was independently dispositive. The record did not establish why the original representatives refused, and the district court noted the substitution without examining it. On remand, the district court is to consider both the substitution and the reasons behind it.

Fourth, parties negotiating equity-based, contingent, or otherwise non-cash settlements should not assume that value uncertainty alone will doom judicial approval. The panel found no inherent problem with an equity stake tied to a defendant’s uncertain future and concluded that the overall terms fell within a reasonable range the district court could approve; it did not itself approve the settlement, which it vacated on other grounds. That still suggests creative settlement structures remain available where a defendant’s balance sheet cannot support a large cash payment. But contingent value has to be valued. The panel’s “one significant concern” was that the district court never made findings on the anticipated value of the cash-demand option because it never examined the defendant’s revenue statements, leaving a material component of the consideration unquantified in the record. Proponents of a contingent structure should put the underlying financial data in front of the court and ask for findings on each payout pathway. And what the court will not tolerate is pairing that creativity with an allocation process that lacks independent representation for the group receiving the least favorable terms.

Fifth, the panel offered a suggestion rather than a directive about the 2018 amendments to Rule 23(e)(2): for future reference and for the sake of consistency, courts “may now, going forward, find it most helpful to frame their analysis in terms of the factors as set forth in Rule 23(e)” instead of the circuit’s older pre-amendment multi-factor list. The court added that “[a]ny difference between the standards is not dispositive in this case.” Read for what it is, that is an invitation, not a new requirement. Even so, settlement proponents may find it prudent to build a record addressing adequacy of representation, arm’s-length negotiation, adequacy of relief, and equitable treatment of class members relative to each other as distinct, independently briefed issues, rather than treating them as a single undifferentiated fairness inquiry.

Conclusion

The Seventh Circuit’s decision leaves the underlying Clearview AI biometric privacy litigation unresolved more than six years after it began, and it sends the parties back to a settlement table where the equity-stake framework may well survive but the allocation process must change. For litigators and in-house counsel tracking biometric and data privacy class actions, the case is less about facial recognition technology than about the mechanics of class settlement design: when a settlement divides a common fund unevenly among subclasses, Rule 23 demands that each side of that division have its own seat at the table before, not after, the deal is struck. As biometric and AI-driven data collection continues to generate large, multi-jurisdictional class exposure, companies should expect that same structural scrutiny to follow any settlement that treats similarly situated class members differently.