Page 2GOOGLE LLC; XXVI HOLDINGS INC; ALPHABET INC.,

Defendants - Appellees .

Appeal from the United States District Court for the Northern District of California Edward J. Davila, District Judge, Presiding

Argued and Submitted June 11, 2025 Pasadena, California

Filed August 26, 2026

Before: Richard R. Clifton, Jay S. Bybee, and Danielle J. Forrest, Circuit Judges. *

Opinion by Judge Clifton; Dissent by Judge Forrest

Page 3SUMMARY **

The panel affirmed the district court's order denying appellants' objections and granting final approval of a settlement agreement reached between plaintiff class representatives and defendants Google LLC and its related entities.

Plaintiffs alleged, on behalf of a class of an estimated 247.7 million persons residing in the United States, that Google violated its users' privacy rights by continuously tracking their location and storing that data even after users had switched off the 'Location History' feature on their devices. The parties submitted a preliminary settlement agreement that provided for injunctive relief and a settlement fund to be distributed to cy pres recipients approved by the district court. Objectors argued that the cy pres provision was improper. The district court held a Fed. R. Civ. P. 23(e)(2) fairness hearing, overruled the objections, and granted final settlement approval. The district court concluded that distribution of settlement funds under the cy pres doctrine was appropriate because a claims process would be too costly and each class members’ pro rata share of the settlement would be de minimis .

The panel rejected Objectors’ challenges to the district court's approval of the settlement agreement. First, the panel rejected Objectors' contention that the district court failed to apply Fed. R. Civ. P. 23(e)(2)(C)(ii)—requiring thePage 4 court to consider the “effectiveness of any proposed method of distributing relief to the class'-in its determination of whether the proposed settlement was fair, reasonable, and adequate. The panel held the district court appropriately considered the enumerated factors under revised Rule 23(e)(2), as amended in 2018. Second, the panel held that the district court satisfied the requirements of Rule 23(e), and did not improperly apply a broad presumption of settlement validity. Third, noting that this court has repeatedly approved cy pres -only distributions, the panel rejected Objectors’ contention contend that the “plain language” of Rule 23(e)(2)(C)(ii) always requires rejection of a cy pres -only settlement because it does not result in any direct distribution to the class. Fourth, the panel affirmed the district court's holding that a direct distribution to all class members was infeasible. Fifth, the panel held that the district court correctly applied the substantial nexus test to determine that the cy pres recipient awards were appropriate as part of a settlement that was fair, adequate, and reasonable.

Finally, the panel declined to consider Objectors’ argument that the settlement and approval order exceeded the district court's Article III authority because Objectors failed to raise the argument to the district court.

Dissenting, Judge Forrest would reverse because she would hold that the district court failed to adhere to important limitations on cy pres disbursement. Its use is appropriate only when settlement funds are truly nondistributable to the absent class members, and that has not been shown in this case.

Page 5Samuel Issacharoff (argued), Samuel Issacharoff, New York, New York; Michael W. Sobol, Jallé H. Dafa, Michael Levin-Gesundheit, Michael K. Sheen, John D. Maher, and Melissa Gardner, Lieff Cabraser Heimann & Bernstein LLP, San Francisco, California; Nicholas Diamand, Lieff Cabraser Heimann & Bernstein LLP, New York, New York; Tina Wolfson, Theodore W. Maya, Bradley K. King and Henry J. Kelston, Ahdoot & Wolfson PC, Burbank, California; for Plaintiffs-Appellees.

Theodore H. Frank (argued) and Anna W. St. John, Hamilton Lincoln Law Institute, Washington, D.C., for Objectors-Appellants.

Benjamin W. Berkowitz (argued), Ian A. Kanig, Nicholas D. Marais, and Thomas E. Gorman, Keker Van Nest & Peters LLP, San Francisco, California; for Defendants-Appellees.

Eric H. Wessan, Solicitor General; Brenna Bird, Iowa Attorney General; Office of the Iowa Attorney General, Des Moines, Iowa; Steve Marshall, Alabama Attorney General, Office of the Alabama Attorney General, Montgomery, Alabama; Chris Carr, Georgia Attorney General, Office of the Georgia Attorney General, Atlanta, Georgia; Treg Taylor, Alaska Attorney General, Office of the Alaska Attorney General, Anchorage, Alaska; Raúl R. Labrador, Idaho Attorney General, Office of the Idaho Attorney General, Boise, Idaho; Tim Griffin, Arkansas Attorney General, Office of the Arkansas Attorney General, Little Rock, Arkansas; Theodore E. Rokita, Indiana Attorney General, Office of the Indiana Attorney General, Indianapolis, Indiana; Ashley Moody, Florida Attorney General, Office of the Florida Attorney General,Page 6 Tallahassee, Florida; Liz Murrill, Louisiana Attorney General, Office of the Louisiana Attorney General, Baton Rouge, Louisiana; Austin Knudsen, Montana Attorney General, Office of the Montana Attorney General, Helena, Montana; Alan Wilson, South Carolina Attorney General, Office of the South Carolina Attorney General, Columbia, South Carolina; Michael T. Hilgers, Nebraska Attorney General, Office of the Nebraska Attorney General, Lincoln, Nebraska; Marty Jackley, South Dakota Attorney General, Office of the South Dakota Attorney General, Pierre, South Dakota; Drew Wrigley, North Dakota Attorney General, Office of the North Dakota Attorney General, Bismarck, North Dakota; Jonathan Skrmetti, Tennessee Attorney General, Office of the Tennessee Attorney General, Nashville, Tennessee; Dave Yost, Ohio Attorney General, Office of the Ohio Attorney General, Columbus, Ohio; Sean D. Reyes, Utah Attorney General, Office of the Utah Attorney General, Salt Lake City, Utah; Gentner Drummond, Oklahoma Attorney General, Office of the Oklahoma Attorney General, Oklahoma City, Oklahoma; Jason S. Miyares, Virginia Attorney General, Office of the Virginia Attorney General, Richmond, Virginia; Patrick Morrisey, West Virginia Attorney General, Office of the West Virginia Attorney General, Charleston, West Virginia; for Amici Curiae Iowa, Alabama, Alaska, Arkansas, Florida, Georgia, Idaho, Indiana, Louisiana, Montana, Nebraska, North Dakota, Ohio, Oklahoma, South Carolina, South Dakota, Tennessee, Utah, Virginia, and West Virginia Attorneys General.

Page 7CLIFTON, Circuit Judge:

Appellants John Andren, Matthew Lilley, and Joseph St. John (collectively, 'Objectors') appeal the district court's order granting final approval of a settlement agreement reached between plaintiff class representatives (“Plaintiffs”) and defendants Google LLC and its related entities ('Google'). Plaintiffs alleged, on behalf of a class of an estimated 247.7 million persons residing in the United States, that Google violated its users' privacy rights by continuously tracking their location and storing that data even after users had switched off the 'Location History' feature on their devices.

After years of motion practice, the parties submitted a preliminary settlement agreement that provided for injunctive relief and a $62 million settlement fund to be distributed to cy pres recipients approved by the district court. “[C]ourts have used cy pres to distribute class-actionsettlement funds not amenable to individual claims or to a meaningful pro rata distribution to a nonprofit charitable organization whose work indirectly benefits the class members and advances the public interest.” Cy pres , Black’s Law Dictionary (12th ed. 2024). Objectors argued, among other things, that the cy pres provision was improper because it was feasible to directly distribute the $62 million settlement fund to class members through a claims process. The district court overruled the objections and granted final settlement approval based on its determination that the settlement agreement was fair, adequate, and reasonable. We affirm.

Page 8Plaintiffs alleged that Google falsely misrepresented to users that by turning off the “Location History” setting on their devices, they could prevent Google from creating and storing a record of their movements.

Initially, the named Plaintiffs individually sued Google. These suits were consolidated into a putative class action asserting a violation of users' privacy rights. Following mediation, the parties reached a settlement, and the district court preliminarily certified the class for settlement purposes only. The settlement class included '[a]ll natural persons residing in the United States who used [Apple and Android] mobile devices and whose location information was stored by Google while ‘Location History’ was disabled.” The class definition captures approximately 247.7 million persons— roughly three-quarters of the United States’ population.¹

The settlement provided two forms of relief for the class. First, without admitting any liability, Google agreed to rectify its allegedly wrongful conduct and refrain from similar conduct in the future. Second, Google agreed to pay $62 million into a settlement fund. The fund would pay Plaintiffs' attorneys' fees and expenses, service awards to the named class representatives, and other administrative costs associated with the settlement. The remainder of the settlement fund would be distributed to charitable organizations 'with a track record of addressing privacy concerns on the Internet.” Before the final approval hearing, the parties proposed 21 cy pres recipients with a Page 9recommended funding amount for each one and provided the district court with the intended recipients’ project proposals for how they would use the settlement funds.

Objectors challenged the proposed settlement. Among other things, they asserted that dispersing funds under the cy pres doctrine before making any attempt to pay class members through a claims process was improper. They also asserted that the proposed cy pres recipients would not serve class interests.

The district court held a Rule 23(e)(2) fairness hearing, and approved the settlement. It concluded that distribution of settlement funds under the cy pres doctrine was appropriate because a claims process would be too costly and each class members' pro rata share of the settlement would be de minimis . Ultimately, the district court awarded $18.6 million in attorneys' fees, $151,756.23 in litigation expenses, and $5,000 to each of the named class representatives in service awards. The remaining $42.64 million was divided in varying amounts between 21 cy pres recipients. Objectors appealed.

'We review a district court's approval of a proposed class action settlement, including a proposed cy pres settlement distribution, for abuse of discretion.' Nachshin v. AOL, LLC , 663 F.3d 1034, 1038 (9th Cir. 2011). Such review is 'extremely limited,' and we 'will affirm if the district court judge applies the proper legal standard and his findings of fact are not clearly erroneous.” In re Mego Fin. Corp. Sec. Litig. , 213 F.3d 454, 458 (9th Cir. 2000). We have jurisdiction pursuant to 28 U.S.C. § 1291. We affirm.

Objectors contend that the district court failed to apply Rule 23(e)(2)(C)(ii)-requiring the court to consider 'the effectiveness of any proposed method of distributing relief to the class”—in its determination of whether the proposed settlement was fair, reasonable, and adequate. Fed. R. Civ. P. 23(e)(2)(C)(ii). We disagree.

Rule 23 establishes the procedure for class action settlements. See Fed. R. Civ. P. 23(e). Prior to 2018, the Rule provided that a “court may approve [a proposed settlement] only after a hearing and on finding that it is fair, reasonable, and adequate.' Fed. R. Civ. P. 23(e)(2) (2017). Courts, including our court, identified factors to be considered by a district court in making this determination. See, e.g. , Staton v. Boeing Co. , 327 F.3d 938, 959 (9th Cir. 2003). In 2018, Rule 23 was amended to provide specific factors that a court must consider in determining whether a proposed settlement met the fair, reasonable, and adequate standard. Fed. R. Civ. P. 23(e)(2) (2018).

The advisory committee’s notes to the 2018 amendment caution that the amendment was not intended to “displace” any previously identified factor, but 'rather to focus the court and the lawyers on the core concerns of procedure and substance that should guide the decision whether to approve the proposal.” Fed. R. Civ. P. 23(e)(2) advisory committee’s note to 2018 amendment. We subsequently clarified that it is not an abuse of discretion for a district court to continue to consider the factors we had previously identified. “[I]t is still appropriate for district courts to consider these factors in [a] holistic assessment of settlement fairness.” McKinneyDrobnis v. Oreshack , 16 F.4th 594, 609 n.4 (9th Cir. 2021); see also Briseño v. Henderson , 998 F.3d 1014, 1026 (9th Cir.

Page 112021). However, because “Congress provided district courts with new instructions . . . that require them to go beyond our precedent,” it is legal error for a district court to omit from its analysis any of the factors provided for in the amended Rule 23. Briseño , 998 F.3d at 1026.

We conclude that the district court appropriately considered the enumerated factors under the revised Rule 23(e)(2). The district court assessed the 'amount offered” in the settlement and concluded that “the monetary and injunctive relief is fair, adequate, and reasonable.' It further concluded that a “ cy pres distribution [is] appropriate in this case” because the settlement fund is “nondistributable' under our precedents, where the 'pro rata recovery per [c]lass [m]ember could be no more than twenty-five cents, assuming the [district court] awards no attorneys’ fees, service awards, or administration costs.” It gave a detailed and reasoned response to Objectors’ contentions regarding the availability of other methods of relief and found that distribution of the settlement funds through alternative methods would be infeasible.

The district court did not abuse its discretion by considering these factors under the framework provided in Churchill Village, L.L.C. v. Gen. Elec. , 361 F.3d 566, 575 (9th Cir. 2004), prior to the 2018 amendments to Rule 23. As noted above, the district court “explored comprehensively all [Rule 23(e)(2)] factors” and “g[ave] a reasoned response to all non-frivolous objections.' In re Apple Inc. Device Performance Litig. , 50 F.4th 769, 782 (9th Cir. 2022) (first alteration in original) (citation omitted). Because we conclude that the district court would conduct the same analysis and reach the same result were we to remand, we affirm. See Agarwal v. Arthur G. McKee & Co. , 644 F.2d 803, 807 n.3 (9th Cir. 1981).

Objectors argue that the district court erred in applying a presumption of fairness in determining whether the settlement was valid. 'Rule 23(e)(2) assumes that a class action settlement is invalid.” Briseño , 998 F.3d at 1030. “It also instructs that '[i]f the proposal would bind class members, the court may approve it only after a hearing and only on finding that it is fair, reasonable, and adequate.'' Id. (alteration in original) (quoting Fed. R. Civ. P. 23(e)(2)).

The district court satisfied the requirements of Rule 23(e). The district court held a fairness hearing, weighed the objections, and considered the relevant Rule 23(e) factors in making its determination. The district court did not improperly apply a broad presumption of settlement validity to the entirety of its analysis.

Objectors contend that the “plain language” of Rule 23(e)(2)(C)(ii) always requires rejection of a cy pres -only settlement 2 because it does not result in any direct distribution to the class.

We have expressly 'reject[ed] the suggestion that a district court may not approve a class-action settlement that provides monetary relief only in the form of cy pres payments to third parties.' In re Google Inc. St. View Elec. Commc’ns Litig. , 21 F.4th 1102, 1113 (9th Cir. 2021) (footnote omitted) (' Google Street View '). Our court has Page 13repeatedly approved cy pres -only distributions, and in doing so, we have evaluated whether cy pres is appropriate and whether the cy pres recipients will serve the interests of the class. E.g. , id. ; Lane v. Facebook, Inc. , 696 F.3d 811 (9th Cir. 2012).

While the dissent does not go so far as to mount a categorical challenge to the use of cy pres -only distributions, it comes close, at 34–35. Contrary to the dissent’s suggestion that it may not be for the courts “to decide that they know better how to serve the class's interests,' that is precisely what Rule 23 and caselaw provides for courts to do. The district court considered the interests of all class members, not just the tiny number who filed objections and not just the tiny fraction of the class that might be expected to file financial claims. It was correct in doing so. The dissent's alternative approach would elevate the noisy few over the interests of all class members.³

Objectors do not address whether the cy pres requirements are met in this case. The district court held that a cy pres -only distribution was appropriate given that the $62 million settlement fund divided by 247.7 million class members would result in a de minimis recovery for each class member, further diminished by attorneys’ fees, expenses, administrative costs, and service awards.

Page 14The district court’s holding accords with our precedents deeming a settlement fund non-distributable where 'each class member’s recovery under a direct distribution would be de minimis .” Lane , 696 F.3d at 825 (holding a direct distribution infeasible where $6.5 million was to be divided among 3.6 million class members)⁴ ; see also Google Street View , 21 F.4th at 1114 (holding a direct distribution infeasible where $13 million was to be divided among 60 million class members). Because the pro rata distribution here would be similarly miniscule, the district court did not abuse its discretion in approving the parties' proposed settlement.

Objectors argue that the interests of the class are better served by distributing tiny amounts of money to a tiny fraction of the class rather than directing the money to initiatives protecting privacy interests more generally. Although dividing the settlement funds among all class members would leave less than 25 cents for each class member, Objectors contend that only a very small percentage of the class would actually seek recovery. Specifically, Objectors suggest that only 1.5% of the class members would submit claims, leaving less than $17 to each Page 15claimant, without considering administrative expenses, attorneys’ fees, or any other costs, all of which would reduce the funds available for distribution to individual claimants. But that leaves out 98.5% of class members.

This argument—that dispersing funds to a fraction of the class is preferable to payments to third parties through cy pres— rests upon the false premise that cy pres distributions do not benefit the class. We do not share that view. Google Street View , 21 F.4th at 1116 (“In upholding the validity of cy pres arrangements, we have repeatedly recognized that class members do benefit—albeit indirectly—from a defendant’s payment of funds to an appropriate third party.”).

In addition, the dissent fails to acknowledge that there is no benefit whatsoever to class members from significantly larger administrative costs inherent in the approach the dissent prefers. To the contrary, increased administrative expenditures subtract from the benefits to class members. The dissent reasons, at 32, that “[b]ecause there will always be costs associated with processing and distributing a class settlement,' undisputed administrative costs of up to $8.2 million should not weigh against a direct distribution. But it does not explain how that diversion of $8.2 million to administrative expenses—and the reduction of funds available for cy pres distributions—is more beneficial to the class as a whole than the improved privacy benefits that are the goal of programs funded by the cy pres distributions. The added administrative spending would benefit professional administrators and other vendors, but not the class members.

The district court also properly considered the fact that even if it were feasible to distribute such a minimal amount to each claimant, there was no way to verify whether Page 16claimants fit the class definition. The district court relied on evidence submitted by Google that its 'data-collection practices and systems make it infeasible to identify the individuals who fit this class definition.” On appeal, Objectors conclusorily assert that class members could selfidentify and opt-in to the class. Critically absent from Objectors' briefs is any proposed methodology to verify each claimant's entitlement to settlement funds once they selfidentify. Self-identification would thus be “pure speculation,” and Objectors have not offered a way to verify the claims “with any degree of probability.” Google Street View , 21 F.4th at 1115.

The dissent concedes, at 29-30, that Objectors have failed to offer any proposed methodology for reliably verifying their claims. Instead, the dissent surmises, at 29, that the submission of a declaration or affidavit by each claimant “might suffice” to verify their claims. This proposal, however, creates a scenario where an individual who might not be an actual class member is able to submit a claim and affidavit, which cannot be independently verified by scientifically sound methods. Payment to such an individual would be of no benefit to the class.

In the alternative, the dissent offers, at 29-30, that “ordinary methods of verification” used in consumer class action settlements, see, e.g. , Briseno v. ConAgra Foods, Inc. , 844 F.3d 1121, 1131 (9th Cir. 2017), are 'sufficient' to protect Google's rights. The dissent, however, has not explained how methods verifying the purchase of physical retail products (i.e., cooking oil in Briseno ) would translate to claims involving location tracking on mobile devices, where Google has already submitted evidence that such verification is not possible. Nor does the dissent account for the further erosion of funds available to benefit class Page 17members by the cost of whatever verification methods might be proposed and employed.

Accordingly, we affirm the district court’s holding that a direct distribution is infeasible.

Objectors argue that the district court abused its discretion by approving cy pres recipients who (1) were “too remote' from the plaintiff class and the underlying data privacy issues; and (2) engaged in ideological work at odds with a material portion of the class. Objectors specifically take issue with nine of the twenty-one total cy pres recipients.⁵ We conclude that the district court correctly applied our substantial nexus test to determine that the cy pres recipient awards were appropriate as part of a settlement that was fair, adequate, and reasonable. See Lane , 696 F.3d at 821.

The substantial nexus test contemplates whether the proposed award: (1) addresses the objectives of the underlying action; (2) targets the plaintiff class; and (3) provides “reasonable certainty that any member will be benefitted.” Nachshin , 663 F.3d at 1040. We agree with the district court that the cy pres recipients had documented their commitment to advocate for the protection of data privacy, establishing a substantial nexus to the underlying interests of the class.

Page 18Objectors also mount geographic challenges as to three of the proposed organizations. Objectors assert that Fordham University Center on Law and Information Policy’s proposal concerning 'local communities and populations in the Global South' failed to benefit the plaintiff class residing exclusively in the United States. We disagree. Fordham did not propose to conduct research targeting the Global South; it proposed to host scholars from the Global South to engage in internet privacy research, which is consistent with the underlying interests of the class. In addition, Objectors argue that the proposed projects of the Internet Archive and Electronic Frontier Foundation are 'too remote' because they target a global audience. However, they do not explain how class members residing in the United States would not also benefit from global initiatives. As such, Objectors' geographic challenges fail.

Finally, Objectors argue that the district court abused its discretion by approving cy pres recipients that engaged in ideological work at odds with a substantial portion of the class. As a preliminary matter, Objectors’ attempt to introduce an additional factor under the substantial nexus test is not supported by our precedents. See Nachshin , 663 F.3d at 1040 (describing the three factors).

Objectors’ argument separately fails because a standard mandating universal approval would be unworkable in practice. Seeing that the class here is estimated to include 247.7 million people, or most residents of the United States, that some number might disagree seems unremarkable. This is consistent with our guidance in Lane , where we held district courts are not required to “select a cy pres recipient that the court or class members would find ideal” because 'such an intrusion into the private parties' negotiations would be improper and disruptive to the settlement process.”Page 19 696 F.3d at 821. As such, we will not disturb the district court’s approval of the settlement where Objectors have not otherwise established that the recipients lack a substantial nexus to the class.

Objectors further argue that the settlement and approval order exceeded the district court's Article III authority because it permits the district court to designate recipients, allocate funds, and monitor the cy pres distribution for several years after the funds are to be distributed. We decline to consider this challenge on appeal because Objectors failed to raise this argument to the district court. See United States v. Greger , 716 F.2d 1275, 1277 (9th Cir. 1983).⁶ The issues raised in this challenge are complex, and Objectors’ arguments, if accepted, would significantly disrupt accepted legal doctrine and practice. In these circumstances, it is prudent to await a more fulsome presentation before opining on the merits of Objectors' constitutional challenge. See Camreta v. Greene , 563 U.S. 692, 705 (2011).

AFFIRMED.

Page 20FORREST, Circuit Judge, dissenting:

Cy pres derives from Norman French and refers to doing something 'as near as' possible. Cy Pres , Black's Law Dictionary (12th ed. 2024); see Rhonda Wasserman, Cy Pres in Class Action Settlements , 88 S. Cal. L. Rev. 97, 114 (2014) (explaining the cy pres comes 'from the Norman French expression, cy près comme possible , which means ‘as near as possible’”). In the class action context, cy pres refers to “the practice of distributing settlement funds not amenable to individual claims or meaningful pro rata distribution to nonprofit organizations whose work is determined to indirectly benefit class members.” Frank v. Gaos , 586 U.S. 485, 490–91 (2019) (per curiam) (citing Cy Pres , Black's Law Dictionary (10th ed. 2014)). The theory is that a charitable distribution “for the aggregate, indirect, prospective benefit of the class' may be the 'next best' alternative when a direct distribution to the class members cannot be accomplished. Nachshin v. AOL, LLC , 663 F.3d 1034, 1038 (9th Cir. 2011) (citations omitted).

Cy pres certainly helps solve the problem of what to do with undistributed and undistributable class settlement funds. But courts disbursing settlement funds to actors different than those for whose benefit the funds were obtained in the first place also raises serious questions related to due process and the proper role of the civil justice system. In recognition of this fact, the law places limits on when this distribution tool can be used and instructs courts to select cy pres recipients who in some way represent and will serve the class members’ interests. Id. at 1039. As we have explained, “[w]hen selection of cy pres beneficiaries is not tethered to the nature of the lawsuit and the interests of the silent class members, the selection process may answer Page 21to the whims and self interests of the parties, their counsel, or the court.' Id. But the limitations are not always followed in practice.

For example, courts have dispersed settlement funds collected on behalf of purchasers of price-fixed NASCAR race souvenirs to charities like the Make-a-Wish Foundation, In re Motorsports Merchandise Antitrust Litigation , 160 F. Supp. 2d 1392, 1395–99 (N.D. Ga. 2001)), settlement funds collected in antitrust litigation to art museums, Superior Beverage Co., Inc. v. Owens-Illinois, Inc. , 827 F. Supp. 477, 485–86 (N.D. Ill. 1993), and settlement funds arising out of a wage dispute in the modeling industry to women's organizations when men made up nearly half the class, Fears v. Wilhelmina Model Agency, Inc. , No. 02 Civ. 4911 (HB), 2007 WL 1944343, at *10-11 (S.D.N.Y. July 5, 2007), vacated on other grounds , 315 Fed. App'x 333 (2d Cir. 2009). See Wasserman, supra , at 119. And sometimes the defendants who pay the settlement have held vested interests in the recipient organizations, all but giving settlement funds back to their source—a practice we have condoned only in limited circumstances. See id. at 121 (discussing Lane v. Facebook, Inc. , 696 F.3d 811, 820–21 (9th Cir. 2012), where an allcy pres distribution set up a charity to be run in part by a one of the defendant’s executives).

In this case, the district court failed to adhere to important limitations on cy pres disbursement. It extended our reasoning in In re Google Inc. Street View Electronic Communications Litigation , 21 F.4th 1102 (9th Cir. 2021), and approved the distribution of over $40 million to various charitable organizations without attempting any distribution to the absent class members where, unlike in Google Street View , it is possible to identify the class members and verify their claims. Relaxing the cy pres standard in this manner Page 22undermines its function as an equitable doctrine, Dennis v. Kellogg Co. , 697 F.3d 858, 865 (9th Cir. 2012), which are meant “to secure justice where it would not be secured by the ordinary and existing processes of law,” Trump v. CASA, Inc. , 606 U.S. 831, 842 (2025) (citation omitted). Cy pres is not properly employed merely because the members of a class may be difficult to identify or are likely to receive only a small sum. Its use is appropriate only when settlement funds are truly 'non-distributable' to the absent class members. E.g. , Lane , 696 F.3d at 819; Google Street View , 21 F.4th at 1115. Because that has not been shown to be the case here, I respectfully dissent.

Settlements in “class actions present unique due process concerns for absent class members.'' In re Bluetooth Headset Prods. Liab. Litig. , 654 F.3d 935, 946 (9th Cir. 2011) (citation omitted). Because class members are not present to advance their own interests, there is an “inherent risk that class counsel may collude with the defendants, tacitly reducing the overall settlement in return for a higher attorney's fee.' Id. (citation omitted). For this reason, Federal Rule of Civil Procedure 23(e) imposes specific procedural requirements for these settlements. See id. Namely, the district court must approve all class settlements, and it may do so only “after a hearing and only on finding that [the settlement] is fair, reasonable, and adequate.” Fed. R. Civ. P. 23(e)(2).

Class settlements also present unique administrability problems, including what to do with unclaimed settlement funds and funds that cannot feasibly be distributed to the class members. One option is to issue “additional pro rata Page 23distributions to those class members who did claim funds.” Google Street View , 21 F.4th at 1110–11. A second “option is to permit such funds to escheat to the government.” Id. at 1110 (citing Hodgson v. YB Quezada , 498 F.2d 5, 6 (9th Cir. 1974); 28 U.S.C. § 2042); see Six (6) Mexican Workers v. Ariz. Citrus Growers , 904 F.2d 1301, 1307 (9th Cir. 1990). Third, 'in exceptional circumstances it may be proper to permit unclaimed sums to revert to the [defendant].” Hodgson , 498 F.2d at 6; see Six (6) Mexican Workers , 904 F.2d at 1307. And finally, a district court may use the cy pres doctrine to distribute settlement funds to charitable organizations that serve the interests of the class. See Google Street View , 21 F.4th at 1111; Six (6) Mexican Workers , 904 F.2d at 1307.

Rather than granting the government or the defendant a windfall, cy pres is intended to serve the class's interests by funding organizations that will work to redress the class’s injury or its causes. See Google Street View , 21 F.4th at 1111 . ; Six (6) Mexican Workers , 904 F.2d at 1308–09. But this “next best” theory only holds true where the charitable recipients have a connection to the class’s interests such that the funds will be used in a manner that serves those interests. Nachshin , 663 F.3d at 1038–39. As we have cautioned, “the cy pres doctrine—unbridled by a driving nexus between the plaintiff class and the cy pres beneficiaries-poses many nascent dangers to the fairness of the distribution process.” Id. at 1038.

The named Plaintiffs in this class action independently sued Google, LLC, and its corporate relatives, alleging that Google tracked mobile-device users' locations and stored that data even after users switched off the “location history”Page 24 setting on their devices. These suits were consolidated into a putative class action asserting violation of users' privacy rights. Eventually, the parties agreed on a settlement, and the district court preliminarily certified the class for settlement purposes only.

The settlement class included '[a]ll natural persons residing in the United States who used one or more mobile devices and whose Location Information was stored by Google while ‘Location History’ was disabled at any time during the Class Period.' This class definition captures approximately 247.7 million Americans-roughly threequarters of the United States’ population.¹

Settlement provided two kinds of relief for the class. First, Google agreed to rectify its allegedly wrongful conduct and refrain from similar conduct in the future. Second, Google agreed to pay $62 million into a settlement fund. The fund would pay Plaintiffs' attorneys' fees and expenses, service awards to the named class representatives, and other administrative costs associated with the settlement. The entire remainder of the settlement fund would be distributed to charitable organizations. There was no provision for direct payment to class members to compensate for their asserted injury.

Objectors John Andren, Matthew Lilley, and Joseph St. John challenged the proposed settlement. Relevant here, they argued that dispersing funds under the cy pres doctrine before making any attempt to pay class members through a claims process was improper. The district court held a Rule 23(e)(2) fairness hearing and approved the settlement. The court concluded that distribution of settlement funds under Page 25the cy pres doctrine was appropriate because a claims process would be too costly and the class members’ pro rata share of the settlement would be de minimis . Ultimately, the district court awarded $18.6 million in attorney fees, $151,756.23 in litigation expenses, and $5,000 in service awards to the named class representatives. The remaining $42.64 million was divided in varying amounts between 21 cy pres recipients. Objectors appealed.

As an initial matter, I agree with the majority that Objectors forfeited their argument that use of an allcy pres 2 settlement exceeded the district court’s Article III authority. Objectors are not arguing that the federal judiciary lacks power to act in this case, only that the district court exceeded the scope of its power. Therefore, forfeiture principles apply. See Reed Elsevier, Inc. v. Muchnick , 559 U.S. 154, 160-61 (2010) (recognizing that an argument about judicial power is only “jurisdictional,” and therefore exempt from forfeiture, when it 'delineat[es] the classes of cases (subject-matter jurisdiction) and the persons (personal jurisdiction) implicating that authority” (citation omitted)). And while we Page 26have discretion to ignore forfeiture, I would not here because Objectors raised their constitutional challenge at the last minute and devoted only a few pages of their briefing to it, defeating the benefits of a full adjudicative process.

With that said, the concern that Objectors raise is not trivial. See Keepseagle v. Perdue , 856 F.3d 1039, 1071 (D.C. Cir. 2017) (Brown, J., dissenting) (asserting that “ cy pres is problematic for judicial power” in relation to justiciability requirements); Klier v. Elf Atochem N. Am., Inc. , 658 F.3d 468, 480-82 (5th Cir. 2011) (Jones, C.J., concurring) (explaining that, while the question of '[w]hether cy pres distributions violate the Constitution or Rules Enabling Act has not . . . been fully litigated in any court,” the doctrine poses 'opportunities for abuse' that justify taking care to 'avoid the legal complications that assuredly arise when judges award surplus settlement funds to charities and civic organizations”). While it is prudent to wait before wading into the constitutional issues that Objectors raise, they are ones that deserve consideration when properly presented.

Federal Rule of Civil Procedure 23 mandates the procedure for class-action settlements, including the factors that courts should consider when determining whether a settlement is “fair, reasonable, and adequate.” See Fed. R. Civ. P. 23(e)(2). Among other things, these factors include whether “the relief provided for the class is adequate, taking into account: . . . the effectiveness of any proposed method of distributing relief to the class, including the method of processing class-member claims.” Fed. R. Civ. P. 23(e)(2)(C)(ii). “We review a district court’s approval of a proposed class action settlement, including a proposed cy Page 27pres settlement distribution, for abuse of discretion.' Google Street View , 21 F.4th at 1110 (citation omitted).

Objectors argue that the district court erred in concluding that the requirements for employing cy pres distribution are met here.³ Namely, they argue that the district court erred in concluding that the settlement fund is “non-distributable.” E.g. , Lane , 696 F.3d at 819; Google Street View , 21 F.4th at 1115. I agree.

A fund is “‘non-distributable’ when 'the proof of individual claims would be burdensome or distribution of damages costly.'' Lane , 696 F.3d at 819 (quoting Nachshin , 663 F.3d at 1038). Put another way, “[i]f it were feasible to distribute the settlement fund to the class members, a cy pres settlement would not be employed.” Google Street View , 21 F.4th at 1115. The district court determined that the monetary settlement here is non-distributable because it is not feasible to identify the class members, a claimsadministration process would be prohibitively expensive, and the class members would receive only de minimis recovery under a claims process. These three justifications—Page 28 alone or together—do not justify depriving class members an opportunity to claim settlement funds secured on their behalf.

The district court relied on Google Street View in concluding that a pure cy pres distribution was proper because Google could not feasibly identify the class members here. At issue in Google Street View was the 'specially adapted cars' that Google launched to capture “panoramic street-level images from numerous points along roads throughout the world.” 21 F.4th at 1107. In addition to pictures, the cars collected information about nearby Wi-Fi networks and “payload data” (“substantive information such as emails, usernames, passwords, videos, photographs, and documents” that were “transmitted over unencrypted Wi-Fi networks' from homes and other locations that the cars passed by). Id. at 1108. While there was no precedent establishing that an allcy pres distribution was proper, we observed that some authority indicates such settlements are “permissible under appropriate circumstances.” Id. at 1111; see also id. at 1113 (rejecting a “blanket rule” against allcy pres settlements). And we held that such circumstances were present “[b]ecause self-identification would be pure speculation, and any meaningful forensic verification of claims would be prohibitively costly and time-consuming.” Id. at 1115.

Google Street View itself was an expansion of cy pres — the wisdom of which could be questioned. See Emily Potucek, Note, Class Action Cy Pres : A Call for Reform Amidst Inconsistency and Criticism , 27 Lewis & Clark L. Rev. 713, 723-24 & n.62 (2023) (observing that, while '[t]he Ninth Circuit . . . purports to use a variation of thePage 29 [same test other circuits use] to determine when cy pres remedies are appropriate,” its willingness to approve allcy pres settlements in cases like Google Street View has been highly irregular). But even accepting its reasoning, as we must, it is distinguishable from this case.

Here, verification of class members is not impossible. Google presented evidence that it could not reliably identify class members because the user-account information that it maintains is not readily correlated to specific users. For example, one person can have more than one Google account, and a single Google account can be used by multiple people. But that should not be the end of the analysis. Reliable self-identification by affected users may be possible here where it was not in Google Street View . While individuals could not know whether a Google car was close to their home and downloading data as it passed by, mobile-device users may know or be able to determine whether their 'Location History' setting was activated during the class period. That is not to say that claim verification would be unnecessary, e.g. , TransUnion LLC v. Ramirez , 594 U.S. 413, 431 (2021) ('Every class member must have Article III standing in order to recover individual damages.”), but an unrebutted declaration or affidavit might suffice, e.g. , Briseno v. ConAgra Foods, Inc. , 844 F.3d 1121, 1132 (9th Cir. 2017), abrogated in part on other grounds , TransUnion , 594 U.S. at 431.

Google insists that verification is impossible because it lacks individualized account data, an assertion the majority uncritically accepts. Maj. Op. at 15-16. But we generally presume that ordinary methods of verification-'claim administrators, various auditing processes, sampling for fraud detection, follow-up notices to explain the claims process, and other techniques tailored by the parties and the Page 30court'-are sufficient to protect the rights of defendants confronted with class-action settlements like this one. Id. at 1131 (citation omitted). Google has not explained why these methods, which have been used effectively in other cases and will be used effectively again, have no application here. Accordingly, the difficulty of identifying the class members is not a valid justification for depriving the unnamed class members of the opportunity to submit their claims, particularly where there has been no attempt to identify them. See Molski v. Gleich , 318 F.3d 937, 955 (9th Cir. 2003) (observing that cy pres is inappropriate where “there is no evidence that proof of individual claims would be burdensome or that distribution of damages would be costly”). We need not insist on a proposed methodology for verification from Objectors, as the majority does, where the usual methodology may do the job. Maj. Op. at 16–17.

The district court reasoned that an allcy pres distribution was appropriate because processing even a low rate of claims would consume an inordinate amount of the class funds—$1.9 million for a 1% claims rate, $4 million for a 3% claims rate, and $8.2 million for a 7% claims rate. One need not second guess these figures to reject this rationale.

Cy pres may be appropriate when the 'distribution of damages [is] costly,'' Lane , 696 F.3d at 819 (citation omitted), but we have not held that this consideration allows courts to approve settlements that forego any effort at class distribution where it otherwise may be feasible. Much of our early caselaw repeated the distribution-cost consideration without applying it to the facts of the given case. See, e.g. , Six (6) Mexican Workers , 904 F.2d at 1305, 1307; Nachshin , 663 F.3d at 1038-41; Dennis , 697 F.3d at 865–68. Only once Page 31have we approved using cy pres solely based on the claimsdistribution cost. See In re Google Referrer Header Priv. Litig. , 869 F.3d 737, 741-43 (9th Cir. 2017), vacated on other grounds , Frank , 586 U.S. 485. And in that case, 'the cost of verifying and ‘sending out very small payments to millions of class members would [have] exceed[ed] the total monetary benefit obtained by the class.’” Id. at 742.

Because Google Referrer was vacated and remanded by the Supreme Court for further proceedings related to standing, it is not binding and its reasoning related to cy pres has value only to the extent that it can persuade. See, e.g. , Roe v. Anderson , 134 F.3d 1400, 1404 (9th Cir. 1998). I am not persuaded. It is one thing to conclude that an allcy pres distribution is warranted when the class fund would be fully exhausted by administering a claims process. Courts should not blind themselves to reality and pursue a claims process that literally will provide no benefit to class members. Cf. Samuel L. Bray & Paul B. Miller, Getting into Equity , 97 Notre Dame L. Rev. 1763, 1784 (2022) (explaining that, “in order to preserve its functionality,' equity has historically resisted being 'reduced to norms and enabling doctrines meant to provide ex ante guidance” in all cases); Henry E. Smith, Equity as Meta-Law , 130 Yale L.J. 1050, 1110 (2021) (“Part of equitable analysis is making sure that equity does not produce inequity. Equitable rules of thumb must be set aside in a case if they produce the kind of injustice equity aims to prevent.'). But that is not the situation here. The settlement fund remaining after payment of attorney fees and other administrative expenses exceeds $40 million. And the highest estimated distribution cost based on an assumed claims rate was $8.2 million. That leaves over $30 million for distribution to the individuals who suffered the injury that the settlement is supposed to remedy.

Page 32Because there will always be costs associated with processing and distributing a class settlement, the cost justification for cy pres must be narrow: costs must be so high as to make effective distribution infeasible.⁴ See Mirfasihi v. Fleet Mortg. Corp. , 356 F.3d 781, 784 (7th Cir. 2004) (Posner, J.) (“In the class action context[,] the reason for appealing to cy pres is to prevent the defendant from walking away from the litigation scot-free because of the infeasibility of distributing the proceeds of the settlement . . . to the class members.'). A standard unconnected to the probable impact on the efficacy of class distribution risks cy pres becoming a first choice, not a “next best” choice. E.g. , Google Street View , 21 F.4th at 1115 (quoting Nachshin , 663 F.3d at 1036). Where the district court made no findings regarding the efficacy of distribution relative to the likely administrative costs, it abused its discretion in relying on this justification for proceeding with a cy pres distribution.

3.

Finally, the district court reasoned that a pure cy pres distribution was appropriate because the pro rata recovery Page 33for each class member would be around 25 cents before deducting fees and costs. Again, the district court misapprehends precedent.

Our earliest reference to the individual recovery amount informing the feasibility of class settlement distribution was in Lane . 696 F.3d at 821. But in that case the Objectors conceded 'that direct monetary payments to the class of remaining settlement funds would be infeasible given that each class member’s direct recovery would be de minimis .” Id. We did not analyze feasibility as informed by the individual class member recovery amount or establish a standard for courts to apply in assessing that issue going forward. See Burbank-Glendale Pasadena Airport Auth. v. City of Burbank , 136 F.3d 1360, 1363 (9th Cir. 1998); see also Henry Campbell Black, Handbook on the Law of Judicial Precedents, or the Science of Case Law 3, 126–28 (1912) (recognizing that a precedent’s strength turns on the degree of elaboration in the opinion). But see United States v. Contreras , 593 F.3d 1135, 1136 (9th Cir. 2010) (en banc) (per curiam) (suggesting that a later panel may be bound in limited circumstances by an earlier panel’s silence).

We dealt with the de miminis -recovery issue more directly in Google Referrer , upholding approval of a cy pres distribution where the settlement for a 129-million-member class was $8.5 million, noting that the resulting pro rata recovery of a few cents was “a de minimis amount if ever there was one.” 869 F.3d at 740, 742. But as noted, this case is no longer binding, and its analysis is flawed. See Roe , 134 F.3d at 1404.

Google Referrer 's first misstep was reading Lane as holding that de minimis recovery renders distribution infeasible when Lane made no such holding. See Google Page 34Referrer , 869 F.3d at 742. Google Referrer also provided no standard for what constitutes ' de minimis ' recovery. Five cents? Twenty-five cents? A dollar? Five dollars? As Objectors stress, it is routine for large class actions to result in low per capita recoveries for class members. See, e.g. , In re Yahoo! Inc. Customer Data Sec. Breach Litig. , No. 16MD-02752-LHK, 2020 WL 4601627, at *1 (N.D. Cal. Aug. 11, 2020) (involving a per capita recovery 'well below' $1.02 to $1.46 per class member); see also McCabe v. Six Continents Hotels, Inc. , No. 12-cv-04818 NC, 2015 WL 3990915, at *9 (N.D. Cal. June 30, 2015) (identifying several class-action settlements with per capita recoveries below $10).

The purpose of equity is to discourage opportunism and other undesirable behaviors by filling the gaps left by law’s default rules. Smith, supra , at 1071–81; see Bray & Miller, supra , at 1784 (“Corrective equity consists in interventions made against, and so adjectivally in relation to, law.”). It is not to give judges discretion to override otherwise applicable rules. See Smith, supra , at 1140 (explaining that, while “discretion is important, even essential,” to equity, it is “not unbridled'). If the de minimis nature of a distribution matters, it must be because it makes distribution infeasible and, therefore, contrary to equity-not because of the de minimis amount of individual distributions in and of itself.

The majority rejects the idea 'that the interests of the class are better served by distributing tiny amounts of money to a tiny fraction of the class rather than directing the money to initiatives protecting privacy interests more generally.' Maj. Op. at 14. But why is it more equitable for an uninjured outside party that can only serve the class's interests indirectly to receive a significant sum rather than the directly injured parties to receive a small sum? And perhaps more Page 35important, where a direct distribution is possible, why should courts get to decide that they know better how to serve the class’s interests?⁵

In both Google Referrer and this case, the harm suffered by the class was unliquidated. Cf. 869 F.3d at 740 (describing the harm to class members as the sharing of “users’ search terms . . . with third parties”). In that circumstance, settlement distribution can be structured to give the class members who submit a claim a higher pro rata share than they would have received based on a full-class distribution. This approach does not risk some class members being overcompensated and it ensures that at least some of the settlement recovery goes to the injured population for which the class action was brought.

Where the class members’ injury is liquidated, we have held that cy pres distribution is justified based on de minimis direct disbursement where there has been a claims process and further distribution of the unclaimed settlement funds would be de minimis . In re Easysaver Rewards Litg. , 906 F.3d 747, 753, 761 (9th Cir. 2018). In Easysaver , we declined the objector’s proposal to disburse additional funds Page 36to the class members who had already received their liquidated share of the recovery because “the district court was under no obligation to adopt a distribution approach that might overcompensate claimants.” Id. at 761. And while it might have been technically feasible to identify nonclaimant class members and distribute the remaining fund to them, it was not an abuse of discretion for the district court to resort to cy pres where the non-claimants would receive the ' de minimis ' amount of less than $3.00. Id. (quoting Lane , 696 F.3d at 821). But in this circumstance, cy pres is being used as a “next best” choice. And this line of precedent is inapplicable where class distribution has not even been attempted.

In sum, we have not held that an allcy pres distribution is justified anytime class members will receive de minimis distributions. And resorting to cy pres after a claims process fails to fully distribute settlement funds presents fewer dueprocess and fairness concerns than entirely replacing class distribution with cy pres distribution. See Nachshin , 663 F.3d at 1039 (observing that cy pres is at its most problematic when there is a serious risk that “the selection process may answer to the whims and self interests of the parties, their counsel, or the court,” rather than to “the nature of the lawsuit and the interests of the silent class members”).

Cy pres is justified when it is the “next best” choice for distributing class funds. Courts must start with the presumption that distributing settlement funds to the class members is the first best choice. This presumption can be overcome, but, given the significant due-process and fairness concerns at play, courts must ensure that direct distribution truly is infeasible. The district court’s approval Page 37of an allcy pres distribution of the monetary settlement secured for the class in this case is incongruent with this legal framework. Thus, I would reverse, and I do not reach whether the cy pres recipients approved by the district court were proper.

What is at stake is more than the distribution of what, I readily admit, is often a very small sum per individual class member. Relaxing the requirements for cy pres distribution threatens to sideline focus on the property rights of those for whose benefit the litigation was brought in the first place, which runs headlong into due-process problems. Cy pres is an extraordinary tool for extraordinary cases. The record does not establish that this is such a case where there has been no attempt at distribution and there is no obvious reason to think that distribution would be ineffectual.

I respectfully dissent.