Jane Doe v. Uber Technologies, Inc.

NOT FOR PUBLICATION FILED UNITED STATES COURT OF APPEALS JAN 13 2025 MOLLY C. DWYER, CLERK U.S. COURT OF APPEALS FOR THE NINTH CIRCUIT

JANE DOE, No. 22-16562

Plaintiff-Appellant, D.C. No. 3:19-cv-03310-JSC

v. AMENDED MEMORANDUM* UBER TECHNOLOGIES, INC.; et al.,

Defendants-Appellees.

Appeal from the United States District Court for the Northern District of California Jacqueline Scott Corley, District Judge, Presiding

Argued and Submitted November 14, 2023 Submission Withdrawn January 9, 2024 Resubmitted August 8, 2024 San Jose, California

Before: GRABER, PAEZ, and FRIEDLAND, Circuit Judges. Partial Concurrence and Partial Dissent by Judge GRABER.

This case involves the sexual assault of a rideshare passenger by an individual posing as an authorized Uber driver. Plaintiff Jane Doe appeals the district court’s dismissal of three tort claims based on ostensible agency against Uber Technologies, Inc., Rasier, LLC, and Rasier CA, LLC (collectively, “Uber”)

* This disposition is not appropriate for publication and is not precedent except as provided by Ninth Circuit Rule 36-3.

as well as the district court’s grant of summary judgment on her negligence claim. We affirm the dismissal of the claims based on ostensible agency and reverse the grant of summary judgment on the negligence claim. 1. The district court did not err by dismissing Plaintiff’s tort claims based on ostensible agency. Even accepting Plaintiff’s allegations as true, Plaintiff has not plausibly alleged that her assailant was acting within the scope of his ostensible employment when he assaulted her. See Lisa M. v. Henry Mayo Newhall Mem’l Hosp., 12 Cal. 4th 291, 299–306 (1995); Daza v. L.A. Cmty. Coll. Dist., 247 Cal. App. 4th 260, 268–69 (2016). Thus, Uber cannot be held vicariously liable for the actions of Plaintiff’s assailant. 2. The district court erred, however, by granting Uber summary judgment on Plaintiff’s negligence claim. In particular, the district court concluded that Uber did not owe Plaintiff a duty of care. See Doe v. Uber Techs., Inc., No. 19-CV-03310-JSC, 2022 WL 4281363, at *4 (N.D. Cal. Sept. 15, 2022). We disagree.

Plaintiff asserts that Uber owes its app users a general duty of ordinary care under California Civil Code Section 1714(a), which establishes the default rule that “each person has a duty to exercise, in his or her activities, reasonable care for the safety of others.” Kuciemba v. Victory Woodworks, Inc., 14 Cal. 5th 993, 939 (2023) (quoting Brown v. USA Taekwondo, 11 Cal. 5th 204, 214 (2021)). The issue here is thus whether Uber owed Plaintiff a general duty of ordinary care *3 because it “create[d] or contribute[d]” to her risk of sexual assault at the hands of an imposter driver.1 Id. at 940.

In a prior order, we certified the question to the California Supreme Court. See Doe v. Uber Techs., Inc., 90 F.4th 946 (9th Cir. 2024). “The . . . court declined our request for certification. Accordingly, we must ‘predict as best we can what the California Supreme Court would do in these circumstances.’” Marin Tug & Barge, Inc. v. Westport Petroleum, Inc., 271 F.3d 825, 829–30 (9th Cir. 2001) (quoting Pacheco v. United States, 220 F.3d 1126, 1131 (9th Cir. 2000)).2 For the reasons below, we believe the California Supreme Court would hold, on the undisputed facts here, that Uber owes its app users a duty to exercise reasonable care regarding their safety, including a “duty not to expose others to an unreasonable risk of injury at the hands of third parties.” Lugtu v. Cal. Highway Patrol, 26 Cal. 4th 703, 717 (2001). 3. As an initial matter, we must address a decision from the California Court of Appeal that Uber argues controls the outcome of this case. In Jane Doe No. 1 v.

1 Plaintiff also asserted that Uber owes its app users a duty of care under a “special relationship” theory of liability. Because we hold that Uber owed Plaintiff a duty of care under a misfeasance theory of liability, we need not reach the “special relationship” question. 2 As we have previously acknowledged, “[t]he California Supreme Court’s denial of our certification request is in no way an expression of its opinion on the correctness of the judgments” motivating that certification request. In re K F Dairies, Inc. & Affiliates, 224 F.3d 922, 925 n.3 (9th Cir. 2000) (citing, among others, Trope v. Katz, 11 Cal. 4th 274, 287 n.1 (1995)).

Uber Techs., Inc., 79 Cal. App. 5th 410 (2022), the plaintiffs—women who had been abducted and sexually assaulted by assailants posing as authorized Uber drivers—brought negligence claims against Uber. The court concluded that Uber did not owe the plaintiffs a duty of care because they failed to “allege[] actions by the Uber entities that created a peril, that is, an unreasonable risk of harm to others.” Id. at 426 (internal quotation marks and citation omitted). The court relied on a “necessary component” test, reasoning that “[t]he violence that harmed the Jane Does—abduction and rape—is not a necessary component of the Uber business model,” so Uber did not owe them a duty of care. Id. at 427 (cleaned up). The California Supreme Court declined to review or depublish the decision.3

“When interpreting state law, we are bound to follow the decisions of the state’s highest court, and when the state supreme court has not spoken on an issue, we must determine what result the court would reach based on state appellate court opinions, statutes and treatises.” Mudpie, Inc. v. Travelers Cas. Ins. Co. of Am., 15

3 The California Supreme Court’s decision in Trope v. Katz, 11 Cal. 4th 274 (1995), affirmatively counsels against treating that court’s denial of review as an acquiescence to the decision on the merits. Our court has nonetheless stated that we do give intermediate state court decisions extra weight when there has been a denial of review. Herrera v. Zumiez, Inc., 953 F.3d 1063, 1069-70 (9th Cir. 2020); Poublon v. C.H. Robinson Co., 846 F.3d 1251, 1267 (9th Cir. 2017). Our articulations of that heightened deference principle have anticipated that there could still be convincing evidence that the state supreme court would disagree with the state intermediate court even after a denial of review. Herrera, 953 F.3d at 1069-70; Poublon, 846 F.3d at 1266-67. We believe that is the situation here, for the reasons explained.

F.4th 885, 889 (9th Cir. 2021) (quoting Diaz v. Kubler Corp., 785 F.3d 1326, 1329 (9th Cir. 2015)). “Decisions of the California Supreme Court, including reasoned dicta, are binding on us as to California law.” Muniz v. United Parcel Serv., Inc., 738 F.3d 214, 219 (9th Cir. 2013). By contrast, decisions of the California Court of Appeal “do not bind each other or us,” id., but are especially persuasive “where, as in this case, the highest court has refused to review the lower court’s decision.” Herrera, 953 F.3d at 1069. “[I]n the absence of convincing evidence that the highest court of the state would decide differently, a federal court is obligated to follow the decisions of the state’s intermediate courts.” In re Kirkland, 915 F.2d 1236, 1239 (9th Cir. 1990) (internal quotation marks and citation omitted).

In this case, there is “convincing evidence” that the California Supreme Court would not follow Jane Doe No. 1.4 Specifically, we conclude that Jane Doe No. 1’s “necessary component” test is irreconcilable with the California Supreme Court’s own pronouncements, most recently its decision in Kuciemba v. Victory Woodworks, Inc., 14 Cal. 5th 993 (2023). Cf. In re K F Dairies, 224 F.3d at 925 (disregarding two decisions from the California Court of Appeal that were

4 The dissent mischaracterizes our decision as based merely on “disagreement with the California Court of Appeal’s reasoning.” Dissent at 6. This is incorrect. As we explain, Jane Doe No. 1 is irreconcilable with the California Supreme Court’s own pronouncements, and we believe there is convincing evidence that the California Supreme Court would not follow that decision.

“directly on point” because those cases were “in conflict with generally established principles . . . as articulated by the California Supreme Court”).

In Kuciemba, the defendant had violated a county health order by transferring employees who had been exposed to COVID-19 to a different worksite. 14 Cal. 5th at 1017. As a result, an employee contracted COVID-19, which he later transmitted to his wife. Id. The relevant question there was whether the defendant owed the employee’s wife a duty of care under Section 1714(a) because it had created or contributed to her risk of harm. Id. at 1016-17. The Kuciemba court concluded that the defendant did owe the employee’s wife a duty of care because it had “created a risk of harm by violating a county health order designed to limit the spread of COVID-19.” Id. at 1018. Thus, the court determined that the plaintiffs’ allegations “raise[d] a claim that [the defendant] violated its obligation ‘to exercise due care in [its] own actions so as not to create an unreasonable risk of injury to others.’” Id. (quoting Lugtu, 26 Cal. 4th at 716).

This analysis plainly conflicts with the “necessary component” test applied in Jane Doe No. 1—indeed, applying the “necessary component” test would have precluded the outcome in Kuciemba and other state court decisions. Although the plaintiffs in Kuciemba alleged that the defendant had created the wife’s risk of harm and that a third party “was the conduit” for that risk of harm, the California Supreme Court did not apply a “necessary component” test to the defendant’s *7 conduct. Id at 1017. Nor was there any reason to think in Kuciemba that the defendant’s alleged negligent conduct—exposing its workers to COVID-19—was a “necessary component” of its business model. Instead, the court expressly stated that the “proper question . . . [was] whether the defendant’s ‘entire conduct created a risk of harm’ to the plaintiff.” Id. (quoting Brown, 11 Cal. 5th at 215 n.6).

That Jane Doe No. 1 is plainly inconsistent with Kuciemba is further borne out by the California Supreme Court’s earlier pronouncements. For example, in Lugtu, the court held that an officer owed a duty of ordinary care “to exercise his or her authority in a manner that does not expose such persons to an unreasonable risk of harm” when pulling a vehicle over. 26 Cal. 4th at 707. The Lugtu plaintiffs alleged that the defendant’s affirmative conduct “in directing [them] to stop [their car] in the center median of the freeway, placed plaintiffs [who were subsequently hit by a negligent third-party driver] in a dangerous position and created a serious risk of harm to which they otherwise would not have been exposed.” Id. at 717. At no point, however, did the court require that the third party’s negligent conduct (reckless driving) be a “necessary component” of the defendant’s conduct to hold that the defendant owed the plaintiffs a duty of care—applying such a rule would have undermined the court’s conclusion. See also Weirum v. RKO Gen., Inc., 15 Cal. 3d 40, 46–49 (1975). Notably, the only California court to have considered *8 the decision at all has declared the test’s application unconvincing. See Hacala v. Bird Rides, Inc., 90 Cal. App. 5th 292, 318 n.11 (2023).

Nor can the “necessary component” test be justified in the specific context of third-party criminal conduct. Indeed, California courts have long explained that “[i]f the realizable likelihood that a third person may act in a particular manner is . . . one of the hazards which makes the actor negligent, such an act whether innocent, negligent, intentionally tortious or criminal does not prevent the actor from being liable for harm caused thereby.” Richardson v. Ham, 44 Cal. 2d 772, 777 (1955) (quoting Restatement (First) of Torts § 449 (1934)) (holding that defendants who left bulldozers unattended owed a duty of reasonable care even where injuries were caused by third-party trespassers who drunkenly raced the bulldozers).

To be sure, “[w]e do not lightly reject the view of an intermediate state appellate court on a question of state law, but our responsibility is to decide the case as would the California Supreme Court.” Hunter v. Ayers, 336 F.3d 1007, 1012 (9th Cir. 2003). For the foregoing reasons, we conclude that the California Supreme Court would not follow Jane Doe No. 1.5 We turn next to how we believe the California Supreme Court would decide this case in the first instance.

5 The dissent suggests that the California Supreme Court recharacterized Brown and Regents in Kuciemba. See Dissent at 7–8. This is incorrect for the

4. Based on the California Supreme Court’s own pronouncements, we believe the court would, given the material facts in this case, determine that Uber owes Plaintiff a duty of care.

To briefly summarize, Section 1714(a) of the California Civil Code provides the general rule of duty in California. To determine whether a defendant owes a plaintiff a duty of care under Section 1714(a), courts undertake a two-step inquiry. First, courts determine “whether the defendant’s ‘entire conduct created a risk of harm’ to the plaintiff.” Kuciemba, 14 Cal. 5th at 1017 (quoting Brown, 11 Cal. 5th at 215 n.6). If Section 1714(a) applies, courts then ask whether the resulting duty ought be narrowed or excepted based upon an analysis of the factors delineated in Rowland v. Christian, 69 Cal. 2d 108 (1968). Id. at 1021.

The first step of this inquiry is relatively straightforward: courts must determine whether the plaintiff’s cause of action is “based upon [a] claim that [the defendant’s] affirmative conduct itself . . . placed plaintiffs in a dangerous position and created a serious risk of harm to which they otherwise would not have been

simple reason that the plaintiffs in those cases never raised misfeasance claims. See Petitioners’ Opening Brief on the Merits at 37–64, Brown v. USA Taekwondo, 11 Cal. 5th 204 (2020) (No. S259216); Petitioner’s Opening Brief on the Merits, Regents of Univ. California v. Superior Court, 4 Cal. 5th 607 (2018) (No. S230568), 2016 WL 1168050, at *29–43. We decline to read Kuciemba’s passing mention of previous California Supreme Court decisions as foreclosing legal theories that the plaintiffs in those cases never raised and that, as the dissent acknowledges, the court never examined.

exposed.” Lugtu, 26 Cal. 4th at 717; see also Weirum, 15 Cal. 3d at 49; Kuciemba, 14 Cal. 5th at 1017–18.6 In other words, courts look to the nature of the plaintiff’s “theory of liability” to determine whether they have properly raised such a claim. Brown, 11 Cal. 5th at 219 n.8; see also Kuciemba, 14 Cal. 5th at 1015 n.7. If so, the plaintiff’s “action . . . is based upon a claim of misfeasance, not nonfeasance,” Lugtu, 26 Cal. 4th at 717–18, and “the question of duty is governed by the standards of ordinary care,” Weirum, 15 Cal. 3d at 49.

In light of the foregoing principles, Plaintiff has sufficiently presented a claim that Uber’s “entire conduct created [her] risk of harm.” Kuciemba, 14 Cal. 5th at 1017 (quoting Brown, 11 Cal. 5th at 215 n.6). Plaintiff’s arguments are based entirely on Uber’s own affirmative conduct in creating or contributing to the risk of sexual assault at the hands of a third party in how it conducts its business. The risk at issue here would simply not have existed without Uber’s own conduct, as is evident from Plaintiff’s articulation of the risk of harm in this case—that is, “the risk that an Uber rider may be kidnapped and sexually assaulted by a predator posing as an Uber driver.” Finally, Plaintiff has presented evidence which, when

6 Indeed, the California Supreme Court has repeatedly warned that courts should not examine case-specific facts during the duty inquiry, as that would intrude upon the province of the jury. See, e.g., Cabral v. Ralphs Grocery Co., 51 Cal. 4th 764, 771–74 (2011). Nor should courts consider foreseeability or the closeness of the connection between the defendant’s conduct and the plaintiff’s harm at the first step, as those are Rowland factors. See Hacala, 90 Cal. App. 5th at 318 n.11.

viewed in the light most favorable to her, supports her theory. Thus, Uber owed Plaintiff a duty to use due care in its operations.

In response, Uber primarily relies on Melton v. Boustred, 183 Cal. App. 4th 521 (2010), and Sakiyama v. AMF Bowling Centers, Inc., 110 Cal. App. 4th 398 (2003). Neither decision warrants a different conclusion.

In Melton, the court considered whether the plaintiffs, partygoers who were assaulted by a group of unknown individuals upon their arrival at a party, could maintain a negligence claim against the party’s organizer, who had used social media to advertise the party with an open invitation. 183 Cal. App. at 527. The court ultimately rejected the plaintiffs’ misfeasance theory of liability on the basis that the defendant “did not engage in any active conduct that increased the risk of harm to plaintiffs.” Id. at 533.

Melton, however, is wholly distinguishable. In Melton, the defendant had “merely invited people—including unknown individuals—to attend a party at his house.” Id. at 535. Here, by contrast, Plaintiff’s theory of Uber’s affirmative conduct contributing to her risk of harm is far more involved, including (1) a novel business model that placed strangers together in a situation where one individual had control over the other’s freedom; (2) active encouragement of vulnerable users by Uber through marketing and the extension of novel services like Uber’s remote ordering capability; (3) reliance by app users on Uber’s promotion of safety and its *12 decals, which it did not require terminated drivers to return; and (4) previous incidents involving imposter drivers that had reached “crisis levels.” See also Reynolds v. Pope, No. A155406, 2020 WL 4333532, at *5 (Cal. Ct. App. July 28, 2020).

Turning to Sakiyama, the California Court of Appeal there determined that the owner of a facility that hosted an all-night party “did not have a duty not to allow its facility to be used for such a party, even if it knew or could assume that drugs would be used by some of the attendees.” 110 Cal. App. 4th at 402. The plaintiffs in Sakiyama were injured partygoers and the parents of deceased partygoers who had attended the party, consumed drugs, and then crashed their vehicle on the way home. Id. at 403. Uber argues Sakiyama demonstrates that California courts have consistently refused to treat “commonplace commercial activities that provide an opportunity for, and thus theoretically increase the risk of, criminal conduct by third parties” as misfeasance.

This argument, too, is unavailing. First, the passage cited from Sakiyama took place at the second step of the duty analysis—that is, as part of the weighing of the Rowland factors. See id. at 409. Thus, it does not shed any light on the first step of the duty inquiry. Cf. Hacala, 90 Cal. App. 5th at 318 n.11. Second, Sakiyama itself does not suggest that a broad limitation for “commonplace commercial activities” exists. The “commonplace commercial activities” *13 mentioned in Sakiyama all involved the furnishing of alcoholic beverages, a context with a very specific history that reduces its utility in predicting how the California Supreme Court would resolve the particular duty question here.7 See 110 Cal. App. 4th at 409. 5. The determination that a duty of care exists, however, does not end our analysis. We must also determine whether that duty ought be narrowed or excepted due to “compelling policy considerations,” as determined upon analysis of the Rowland factors. Kuciemba, 14 Cal. 5th at 1021. This analysis “involves the balancing of a number of considerations” identified by the California Supreme Court. Rowland, 69 Cal. 2d at 112–13. We conclude these considerations do not warrant narrowing or excepting the duty recognized here.

We begin with foreseeability, “[t]he most important factor to consider,” and have little problem determining that Plaintiff’s risk of harm was foreseeable. Kesner v. Superior Ct., 1 Cal. 5th 1132, 1145 (2016). Indeed, Plaintiff has identified numerous examples of similar incidents that occurred prior to her assault as well as evidence that Uber was aware of these incidents. Cf. Regents of Univ. of Cal. v. Superior Ct., 4 Cal. 5th 607, 629–30 (2018); Brown, 40 Cal. App. 5th at 1097. Even the Jane Doe No. 1 court concluded that the plaintiffs’ harm in that

7 This history is described in Ennabe v. Manosa, 58 Cal. 4th 697, 705–10 (2014).

case was foreseeable. See 79 Cal. App. 5th at 415, 427. We thus conclude that foreseeability does not weigh in favor of narrowing Uber’s duty to its app users.8

We next consider “the extent of the burden to the defendant and consequences to the community of imposing a duty to exercise care with resulting liability for breach.” Kuciemba, 14 Cal. 5th at 1027 (quoting Rowland, 69 Cal. 2d at 113).9 On this point, Plaintiff has produced evidence that “Uber already has systems in place to develop safety measures and protect its riders.” The fact that a defendant has already managed to focus attention and resources on enhanced safety measures has been sufficient for the California Supreme Court to conclude that the burden policy factor does not warrant narrowing a duty of care. See, e.g., Regents, 4 Cal. 5th at 633. We thus conclude that any burden caused to Uber by recognizing a tort duty also does not weigh in favor of narrowing that duty.

Finally, none of the remaining factors—that is, “moral blame” or the “overall policy of preventing future harm”—justifies carving out an exception to

8 We likewise conclude that the other foreseeability factors—that is, “the degree of certainty that the plaintiff suffered injury” and the “closeness of the connection between the defendant’s conduct and the injury”—do not change the analysis. Rowland, 69 Cal. 2d at 113. 9 Uber suggests that courts applying the Rowland factors must first “identify the specific action or actions the plaintiff claims the defendant had a duty to undertake.” Castaneda v. Olsher, 41 Cal. 4th 1205, 1214 (2007). However, the California Supreme Court has not required this showing outside the premises liability context. Compare id., with Kuciemba, 14 Cal. 5th at 1027–30, and Regents, 4 Cal. 5th at 633. We likewise decline to apply this requirement.

the duty Plaintiff here seeks to recognize.10 Id. at 631–32. With respect to the former, the California Supreme Court has regularly assigned moral blame in “instances where the plaintiffs are particularly powerless or unsophisticated compared to the defendants or where the defendants exercised greater control over the risks at issue.” Id. at 631 (quoting Kesner, 1 Cal. 5th at 1151). As Plaintiff argues, and as Uber does not challenge, “[t]hat is the case here: compared to its riders, Uber has access to more information about potential dangers facing its riders and a superior ability to implement appropriate safeguards.” With respect to the latter, “[t]he overall policy of preventing future harm is ordinarily served, in tort law, by imposing the costs of negligent conduct upon those responsible.” Id. at 632 (quoting Cabral, 51 Cal. 4th at 781). At minimum, the prevention of future harm factor counsels in favor of imposing a duty of reasonable care upon Uber, particularly if, as observed in Regents, “such steps can avert violent episodes like the one that occurred here.” Id. 6. In sum, we conclude that Uber owed Plaintiff a duty to exercise reasonable care regarding her safety. As the California Supreme Court has made clear, this “general duty of due care includes a duty not to expose others to an unreasonable risk of injury at the hands of third parties.” Lugtu, 26 Cal. 4th at 717. To be sure,

10 Neither party offers evidence that “the availability . . . of insurance for the risk involved” would change the analysis. Kuciemba, 14 Cal. 5th at 1021 (quoting Rowland, 69 Cal. 2d at 113).

our conclusion in no way suggests that Uber breached its duty of care, or that the alleged breach caused Plaintiff’s injuries, as those questions lie squarely in the province of the jury. We decide only that Uber owed Plaintiff a duty to exercise reasonable care for her safety, and we correspondingly remand for further proceedings consistent with this disposition.

AFFIRMED IN PART, REVERSED IN PART, AND REMANDED.

Plaintiff shall recover her costs on appeal.

FILED Jane Doe v. Uber Technologies, Inc., No. 22-16562 JAN 13 2025 GRABER, Circuit Judge, concurring in part and dissenting in part: MOLLY C. DWYER, CLERK U.S. COURT OF APPEALS

I concur in Part 1 of the disposition but otherwise respectfully dissent. The principles of federalism required by our precedents mandate that we follow the holding of Jane Doe No. 1 v. Uber Techs. Inc., 294 Cal. Rptr. 3d 664 (Ct. App. 2022), review denied (Aug. 24, 2022), a decision that is consistent with state law as announced by the California Supreme Court.

1. We must apply “an announcement of the state law by an intermediate

appellate court in California in a ruling which apparently has not been disapproved, and there is no convincing evidence that the law of the State is otherwise.” Six Cos. of Cal. v. Joint Highway Dist. No. 13, 311 U.S. 180, 188 (1940); see also Stoner v. N.Y Life Ins. Co., 311 U.S. 464, 467 (1940) (same); West v. Am. Tel. & Tel. Co., 311 U.S. 223, 237 (1940) (same). In other words, the burden is to demonstrate that the state supreme court would not follow the intermediate court’s decision. We are “obligated to follow the decisions of the state’s intermediate appellate courts” unless there is “convincing evidence” that the state’s supreme court would decide differently. Beeman v. Anthem Prescription Mgmt., LLC, 689 F.3d 1002, 1008 (9th Cir. 2012) (en banc) (citation and internal quotation marks omitted).

Our precedents emphasize the importance of following an intermediate *2 appellate state court’s decision where, as here, the state supreme court declines to review that decision. We held in State Farm Fire & Cas. Co. v. Abraio, 874 F.2d 619 (9th Cir. 1989), that the rule requiring us to follow an intermediate appellate state court’s decision “is especially true when the supreme court has refused to review the lower court’s decision.” Id. at 621. Abraio relied on an even earlier case, Tenneco West, Inc. v. Marathon Oil Co., 756 F.2d 769, 771 (9th Cir. 1985). We made the same point in Poublon v. C.H. Robinson Co., 846 F.3d 1251, 1266– 67 (9th Cir. 2017) (noting that the importance of relying on a California Court of Appeal case was heightened when the California Supreme Court had declined review), and in Miller v. County of Santa Cruz, 39 F.3d 1030, 1036 n.5 (9th Cir. 1994), as amended (Dec. 27, 1994) (relying on Hicks v. Feiock, 485 U.S. 624, 630 (1988) to demand “persuasive data” that the state supreme court would decide otherwise), and have made it again in Franklin v. Cmty. Reg’l Med. Ctr., 998 F.3d 867, 871 (9th Cir. 2021).

Furthermore, our precedents suggest that we must follow the holding or result of a state court case even if we disagree with its internal reasoning. In Abraio, for example, we stated that we must find convincing evidence that the California Supreme Court would “decide differently.” Abraio, 874 F.2d at 621 (emphasis added). Similarly, in Poublon we quoted an earlier case that we must be convinced by persuasive data that the state supreme court would “decide *3 otherwise.” Poublon, 846 F.3d at 1266 (emphasis added) (quoting Miller, 39 F.3d at 1036 n.5).

Logic points in the same direction. Suppose that the California Court of Appeal definitively construed an ambiguous state statute but provided no reasoning, and that the California Supreme Court denied review without comment. Despite the lack of reasoning, we would be obligated to follow the holding of the California Court of Appeal in that hypothetical case, unless convinced by persuasive data that the California Supreme Court would reverse. That is true even if we thought there was a better reading of the ambiguous statute. Similarly, imperfect reasoning is not, without more, a ground for refusing to apply a holding of the California Court of Appeal.

In light of our precedents, the majority disposition errs in three respects. First, it places unwarranted reliance on a footnote in K F Dairies, Inc. & Affiliates v. Fireman’s Fund Ins. Co. (In re K F Dairies, Inc. & Affiliates), 224 F.3d 922 (9th Cir. 2000). See id. at 925 n.3 (“The California Supreme Court’s denial of our certification request is in no way an expression of its opinion on the correctness of the judgments” in two intermediate appellate state court cases. (emphasis added)). Contrary to the majority disposition’s contentions, that case does not detract from our caselaw that both pre- and post-dates K F Dairies and requires us to give weight to the California Supreme Court’s earlier denial of review of Jane Doe No. *4 1. A three-judge panel, as we know, cannot overrule a previous three-judge panel. Miller v. Gammie, 335 F.3d 889, 899–900 (9th Cir. 2003) (en banc).

And even if K F Dairies permissibly ruled that the evidence showed convincingly that the California Supreme Court would not follow the particular California Court of Appeal opinion at issue there, its statement of the law on this point is wrong. K F Dairies relied entirely on California state cases holding that the California Supreme Court’s denial of review does not necessarily suggest agreement with an opinion of the intermediate state court, plus one case from our court holding only that denial of review is not a decision on the merits for the purpose of analyzing claim preclusion. K F Dairies, 224 F.3d at 925 n.3. That is, K F Dairies ignored the solid wall of Supreme Court and Ninth Circuit precedent, detailed above, concerning our Erie role.

Moreover, I question the applicability of the footnote in the context here: the California Supreme Court already denied direct review and considered arguments on the merits of Jane Doe No. 1 before declining our request for certification. The California Supreme Court denied the petition to review Jane Doe No. 1 and then rebuffed our request for certification, despite our pointing out potential flaws in Jane Doe No. 1’s reasoning and laying out a path for possible reversal. Where, as here, the state supreme court refuses review not once, but twice, the weight of its refusal––and the already heightened importance of *5 adhering to existing state appellate court precedent––increases commensurately.

Second, the majority disposition ignores the distinct possibility that the California Supreme Court refused to certify our question because it assumed that we would follow Jane Do No. 1, as required by our precedents. A common reason for a state court of last resort to decline certification is the existence of on-point precedent in the state’s intermediate court of appeals, which in general we must apply. Other reasons include that the question (such as an evidentiary issue) does not definitively resolve a claim, or a judgment by the state court that the issue is not important. In this instance such grounds do not seem to be germane so, even though the California Supreme Court did not state a reason for its denial, by process of elimination it would appear that the California Supreme Court is not troubled by our following Jane Doe No. 1. Our certification order laid out all the reasons why some of us had doubts about the California Supreme Court’s willingness to follow Jane Doe No. 1. Nonetheless only one justice voted to grant our “cert petition.” Additionally, the parties’ filings on the certification question largely tracked the issues we raised in our request, and therefore provide no separate reason for the court’s denial.

Third, disagreement with Jane Doe No. 1’s reasoning seems to be the gravamen of the majority disposition. The disposition interprets the “necessary component” test to be irreconcilable with the California Supreme Court’s *6 pronouncements of misfeasance liability. But as discussed above, the majority disposition’s disagreement with the California Court of Appeal’s reasoning is not, without more, a sufficient ground for rejecting its holding in favor of our own.

In sum, our precedents oblige us to follow the California Court of Appeal decision in Jane Doe No. 1. Jane Doe No. 1 held that Uber did not owe a duty of care to individuals who had been abducted and sexually assaulted by assailants posing as authorized Uber drivers. 294 Cal. Rptr. 3d at 678–79. Applying Jane Doe No. 1’s holding here squarely forecloses Plaintiff’s negligence claim. I would affirm.

2. Even setting aside the presumption to follow the intermediate appellate state court’s holding, and contrary to the majority disposition’s view, Jane Doe No. 1’s reasoning aligns with California Supreme Court precedent. Jane Doe No. 1’s “necessary component” test is consistent with how the California Supreme Court recently characterized its rejection of liability for organizations that merely provide opportunities for harm caused by third parties, rather than meaningfully create, or contribute to, the risk of harm.

In Kuciemba v. Victory Woodworks, Inc., 531 P.3d 924 (Cal. 2023), the California Supreme Court explained why California Civil Code section 1714 did not apply to the entities at issue in Brown v. USA Taekwondo, 483 P.3d 159 (Cal. 2021), and Regents of University of California v. Superior Court, 413 P.3d 656 *7 (Cal. 2018):

These defendants, a sport’s governing body and a university, did not create or contribute to the risk of sexual abuse or stabbing. For that reason, the default duty rule of Civil Code section 1714 did not apply, and the starting point for our analysis was instead the alternate rule that generally one owes no duty to control the conduct of another, nor to warn those endangered by such conduct.

Kuciemba, 531 P.3d at 940 (emphasis added) (citations and internal quotation marks omitted). The court in Brown and Regents did not reach the question whether the defendants contributed to the risk of third-party crime because it analyzed only the special relationship theory in those cases. See Brown, 483 P.3d at 165 (“We here focus, along with the parties, on another basis for finding an affirmative duty: [the special relationship theory].”); Regents, 413 P.3d at 660 (“[W]e hold that universities have a special relationship with their students . . . .”). Nevertheless, the California Supreme Court’s later characterization of Brown and Regents as cases in which the defendants did not create or contribute to the risk supports Jane Doe No. 1’s reasoning, particularly because the court in Brown clarified that “[a] defendant may have greater involvement in the plaintiff’s activities than a chance spectator yet play no meaningful part in exposing the plaintiff to harm.” Brown, 483 P.3d at 165 n.5 (emphasis added). If neither the entity responsible for coordinating the sporting events that a coach used as opportunities to sexually abuse young athletes nor the university where an enrolled student stabbed another student meaningfully created or contributed to the risk of *8 those crimes, it follows that Uber did not meaningfully create or contribute to the risk, either.

Finally, I remain unpersuaded that Jane Doe No. 1 conflicts with California law to an extent that would allow us to set it aside. As I see it, the California Supreme Court cases that purport to present such a conflict can be distinguished by unique factual circumstances not relevant here, such as a defendant that “engaged in affirmative misconduct by violating a county health order,” Kuciemba, 531 P.3d at 941; a law enforcement officer’s “affirmative conduct” in directing a plaintiff to stop his car in the center median of the freeway, Lugtu v. Cal. Highway Patrol, 28 P.3d 249, 257 (Cal. 2001); and heavy construction vehicles left unattended and without a proper locking device, Richardson v. Ham, 285 P.2d 269, 270–72 (Cal. 1955). After all, one case does not control another when “the kinds of foreseeable intervening conduct by third parties as well as the risks created by such conduct” are “materially different” between cases. Richardson, 285 P.2d at 271. At the very least, even if those cases can be read to support Plaintiff’s position, they offer no convincing evidence that the California Supreme Court would decide Jane Doe No. 1 differently. Therefore, applying Jane Doe No. 1, as we are required to do, I would affirm the summary judgment for Uber.