No. 24-3609

No. 24-3616

Appeal from the United States District Court for the Southern District of Ohio at Columbus. Nos. 24-cv-00606; 24-00699—Chelsey M. Vascura, Magistrate Judge.

Argued: February 6, 2025

Decided and Filed: September 16, 2025

Before: MURPHY, DAVIS, and BLOOMEKATZ, Circuit Judges.

ARGUED: Paul Giorgianni, GIORGIANNI LAW LLC, Columbus, Ohio, for Appellant Donald Strench. Jonathan P. Schneller, O'MELVENY & MYERS LLP, Los Angeles, California, for Page 2Appellee Hyundai Motor America Corporation Service Company. ON BRIEF: Paul Giorgianni, GIORGIANNI LAW LLC, Columbus, Ohio, Terry V. Hummel, SCHIFF & ASSOCIATES, Columbus, Ohio, for both Appellants. Jonathan P. Schneller, O'MELVENY & MYERS LLP, Los Angeles, California, for both Appellees.

BLOOMEKATZ, J., delivered the opinion of the court in which DAVIS, J., joined. MURPHY, J. (pp. 15–25), delivered a separate opinion concurring in the judgment in part and dissenting in part.

BLOOMEKATZ, Circuit Judge. For years, Kia and Hyundai produced cars that lacked a standard anti-theft device found in most other vehicles. So thieves began to target these cars in large numbers, which led to a nationwide surge in car thefts. In each of these consolidated cases, a teenager stole one of these cars and, while joyriding, injured an innocent driver on the road. Donald Strench suffered multiple fractures and severe injuries to his head. Matthew Moshi ultimately died from the accident. Strench and Moshi's estate sued Hyundai and Kia, respectively. They alleged that the companies were liable under the Ohio Product Liability Act because they had manufactured cars with design defects that made them especially susceptible to theft, resulting in Strench’s and Moshi’s tragic injuries. The district court dismissed the claims for lack of proximate causation. In doing so, it relied on a line of Ohio cases that address an individual car owner's liability for injuries arising from their car's theft. We hold that those cases do not control product liability claims against car manufacturers, and that the design defect claims therefore survive. But we reject some of Strench's and Moshi's other claims on alternative grounds. We therefore affirm in part and reverse in part.

Between 2011 and 2021, Kia America, Inc. (Kia) and Hyundai Motor America Corp. (Hyundai) produced millions of cars with design features that made them easy to steal. Two Page 3factors made these cars vulnerable to theft. First , the cars lacked engine immobilizers, which are basic anti-theft devices that prevent cars from starting without the right key. Immobilizers are relatively cheap—about $170 per vehicle—and effective at preventing theft. The auto industry began adopting them as a standard in the 1990s. But Kia and Hyundai were outliers. By 2015, 96% of other cars had immobilizers, compared to only 26% of Kia and Hyundai vehicles.

Second , the physical construction of the relevant Kias and Hyundais compounded their vulnerability to theft. The cars’ ignition assembly broke apart with minimal force, allowing easy access to the ignition switch. And the ignition switch could be activated with household items like a USB cord or pliers. These physical features, combined with the lack of immobilizers or equivalent devices, made it possible for people with little technical skill to steal the cars in under 90 seconds.

Having praised the effectiveness of immobilizers in prior regulatory filings, Kia and Hyundai understood that without them, their cars would be easy to steal. Eventually, thieves figured that out too and began to target the cars at alarming rates. Between 2020 and 2021, Milwaukee, Wisconsin, saw a nearly eightfold jump in stolen Hyundai and Kia vehicles. In 2022, Columbus, Ohio, reported a similar spike in Hyundai and Kia thefts, which almost doubled the city’s overall theft rate from two years before. The numbers climbed even higher when, in 2022, a YouTube user posted a 'how-to' guide for stealing Kia and Hyundai vehicles. The video spread virally on social media and gave rise to the 'Kia Challenge,' a trend in which individuals—most often teenagers—steal Kias and Hyundais for joyriding.

Thefts of Kias and Hyundais have caused serious harms in some cases. The appeals before us provide two examples. In one, an unlicensed 15-year-old stole a Kia and, while fleeing from police, struck and killed Matthew Moshi. In the other, a 16-year-old stole a Hyundai and, while joyriding, collided with and seriously injured Donald Strench.

Moshi’s estate and Strench sued Kia and Hyundai, respectively, for these injuries. They asserted common law claims for negligence and nuisance. They also brought four different claims under the Ohio Product Liability Act (OPLA) for: (1) manufacturing defect, (2) design defect, (3) inadequate warning or instruction, and (4) nonconformance to representation. See Page 4Ohio Rev. Code Ann. §§ 2307.74–.77. Both manufacturers moved to dismiss, and the district court granted their motions. Fed. R. Civ. P. 12(b)(6). The district court held that Strench and Moshi could not establish proximate causation—an essential element of all their claims—as a matter of Ohio law. The court observed that under Ohio precedents, a car owner's failure to secure a car against theft does not render the owner liable for injuries caused by a thief's negligent driving. It then applied those precedents equally to the car manufacturers, absolving them of liability in these cases. Moshi and Strench appealed.

We review de novo a district court's grant of a motion to dismiss. Spurr v. Pope , 936 F.3d 478, 482 (6th Cir. 2019). To survive a motion to dismiss, a complaint must contain enough facts, taken as true, to state a claim that is 'plausible on its face.' Ashcroft v. Iqbal , 556 U.S. 662, 678 (2009) (citation omitted). On appeal, Moshi and Strench abandon their common law claims, so we limit our review to their four OPLA claims. Kia and Hyundai argue those claims fail for two reasons. First , they say the claims lack proximate causation as a matter of law. Second , they contend that three of the four OPLA claims independently fail for other claim-specific reasons. We address each argument in turn.

To state a claim under OPLA, Strench and Moshi must plausibly allege that a product defect proximately caused their injuries. See Sutowski v. Eli Lilly & Co. , 696 N.E.2d 187, 191–92 (Ohio 1998) (discussing Ohio Rev. Code Ann. § 2307.73(A)(2)). Because OPLA incorporates common law principles of proximate causation, we consult both OPLA and common law cases in analyzing proximate causation. See Kurczi v. Eli Lilly & Co. , 113 F.3d 1426, 1432 (6th Cir. 1997).

Under Ohio law, proximate causation requires “some reasonable connection” between a wrongful act and an injury. Queen City Terminals, Inc. v. Gen. Am. Transp. Corp. , 653 N.E.2d 661, 670 (Ohio 1995) (citation omitted). Foreseeability is a key component of the analysis: a defendant is liable for harms that a reasonably prudent person in the defendant’s position should have anticipated. See Strother v. Hutchinson , 423 N.E.2d 467, 470–71 (Ohio 1981) (per curiam).

Page 5An intervening cause—an event that occurs after a defendant’s wrongful conduct—can break the chain of proximate causation in some cases. See Leibreich v. A.J. Refrigeration, Inc. , 617 N.E.2d 1068, 1071 (Ohio 1993). But that depends on whether the intervening cause was foreseeable to the defendant. Queen City Terminals , 653 N.E.2d at 670. If the defendant could have reasonably anticipated the intervention, then the defendant remains liable. Id. Because of its fact-intensive nature, foreseeability rarely lends itself to categorical rules. Strother , 423 N.E.2d at 471 (noting that proximate cause is hard to exactly define). Ohio law thus generally treats it as a jury question. See Leibreich , 617 N.E.2d at 1071.

One exception is the so-called “key-in-the-car” line of cases. Dan B. Dobbs et al., The Law of Torts §§ 200, 209 (2d ed. 2025). Ohio courts have held that car owners who expose their cars to theft (typically by leaving their keys in the ignition) are not responsible, as a matter of law, for injuries resulting from a thief’s negligent operation of their stolen vehicles.

The Ohio Supreme Court first confronted that scenario in Ross v. Nutt . 203 N.E.2d 118 (Ohio 1964). There, after a vehicle owner left his key in his car, a thief stole the car and injured the plaintiff while driving away. Id. at 120. The plaintiff sued the car owner for negligence, but the Ohio Supreme Court rejected the claim based on proximate causation. Id. The court first considered whether the theft itself was foreseeable to the car owner. Citing 'common knowledge” and the low incidence of car theft, it concluded that the car owner could not have foreseen the theft in the first place. Id. But even if the theft were foreseeable, the court held that the car owner could 'not reasonably anticipate' that the thief would 'cause an accident.' Id. The court then grounded this conclusion in broader doctrine, observing that car owners have “very limited liability” even when they permit someone to use their car. Id. at 121. Given car owners’ limited liability for permitted uses, the court declined to make car owners “insurer[s]” “for acts of a nonpermissive user, a thief.” Id.

Two decades later, in Pendrey v. Barnes , the Ohio Supreme Court reaffirmed Ross on nearly identical facts. 479 N.E.2d 283, 284 (Ohio 1985) (per curiam). Unlike in Ross , in arguing that the wreck was foreseeable, the plaintiff in Pendrey cited government statistics that showed stolen cars 'have a tendency to become involved in accidents.' Id. But the court rejected the plaintiff’s reliance on those statistics as a matter of policy. It reasoned that basing Page 6liability on such general statistics would make “every” car owner liable for a thief’s actions just because the owner left their key in a car. Id. And it added that creating such strict liability was for the legislature, not the courts. Id. The court thus reaffirmed that leaving a key in a car does not, as a matter of law, make theft-related accidents foreseeable to a car owner.²

Since then, Ohio courts have adhered to Ross and Pendrey in suits that have sought to hold a car owner liable for theft-related injuries based only on the owner’s failure to secure a car against theft. See, e.g. , Magill v. Moore , No. CA2005-08-211, 2006 WL 1843627, at *2–3 (Ohio Ct. App. July 3, 2006). And they have extended Ross and Pendrey to some—but not all—suits against construction companies and auto dealerships. Compare Tilton v. Austintown Motors, Inc. , No. 96-CO-71, 1997 WL 816526, at *1-3 (Ohio Ct. App. Dec. 30, 1997), and Lagowski v. Shelly & Sands, Inc. , 38 N.E.3d 456, 461 (Ohio Ct. App. 2015), with Fletcher Trucking v. Columbus Fair Auto Auction, Inc. , Nos. 94APE09-1394/94APE10-1503, 1995 WL 360323, at *6 (Ohio Ct. App. June 13, 1995).

The question we must answer is whether Ross and Pendrey ’s holding about individual car owners extends to OPLA claims against manufacturers. The district court agreed with Kia and Hyundai that they do. But we think that Strench and Moshi have the better of the arguments. That is because when manufacturers are involved, Ohio common law and OPLA both assess foreseeability from the perspective of a manufacturer. Ross and Pendrey , by contrast, assess foreseeability from the perspective of an ordinary car owner, making them not predictive of how the Ohio Supreme Court would evaluate proximate causation here. See Pennington v. State Farm Mut. Auto. Ins. Co. , 553 F.3d 447, 450 (6th Cir. 2009).

Ohio common law measures foreseeability from the perspective of a reasonable person in the defendant's position. See Strother , 423 N.E.2d at 471. That means Ohio courts consider what the particular defendant 'knew' or 'should have known' at the time it acted . Jeffers v. Olexo , 539 N.E.2d 614, 617-18 (Ohio 1989). If the defendant lacks special awareness of a Page 7risk—as in Ross —Ohio courts rely on “common knowledge” and everyday experience to assess foreseeability. Ross , 203 N.E.2d at 120; accord Taylor v. Webster , 231 N.E.2d 870, 873 (Ohio 1967). But if the defendant is a manufacturer, Ohio courts consider whether the manufacturer had special knowledge or expertise that should have made it aware of certain risks, including third-party intervention. See Leibreich , 617 N.E.2d at 1071–72 (jury could find that pedestrians’ injuries stemming from truck driver's actions were foreseeable to manufacturer given its specialized knowledge of truck’s mechanics and intended use); Queen City Terminals , 653 N.E.2d at 671 (reasonable minds could have found intervening act foreseeable given manufacturer’s particular knowledge).

Beyond these common law principles, OPLA also views foreseeability through the eyes of a manufacturer. The statute incorporates foreseeability into its definition of a design defect: a design is defective if its 'foreseeable risks' outweigh its benefits. Ohio Rev. Code Ann. § 2307.75(A). Foreseeable risks, in turn, are the risks that a manufacturer “should recognize” when exercising '[t]he attention, perception, memory, knowledge, and intelligence [of] a reasonable manufacturer.” Id. § 2307.71(A)(6)(b)(i). If a particular manufacturer has 'superior' expertise, then OPLA accounts for that too. Id. § 2307.71(A)(6)(b)(ii).

Although these OPLA provisions define design defects, they also inform its proximate cause analysis. Consider intervening causes in particular. If a third-party’s intervening act-'whether innocent, negligent, intentionally tortious, or criminal'-is 'one of the hazards which makes the [defendant] negligent,' then the act does not sever proximate causation. Fed. Steel & Wire Corp. v. Ruhlin Constr. Co. , 543 N.E.2d 769, 775 (Ohio 1989) (quoting Restatement (Second) of Torts § 449 (1965) (discussing superseding causes)). In Ross and Pendrey , a thief’s careless driving was not one of the foreseeable hazards that made the car owners negligent, so the owners were not liable. But OPLA defines the relevant risks as those that a sophisticated manufacturer should recognize. Ohio Rev. Code Ann. § 2307.71(A)(6)(b). From that standpoint, theft-related accidents may well count among the hazards that make a car's design defective. See Cincinnati v. Beretta U.S.A. Corp. , 768 N.E.2d 1136, 1145-47, 1147 n.7 (Ohio 2002) (allegations that manufacturer failed to “incorporate feasible safety devices that would prevent unauthorized use” of potentially dangerous product can state OPLA claim);Page 8 Perkins v. Wilkinson Sword, Inc. , 700 N.E.2d 1247, 1252 (Ohio 1998). And if so, then theft-related accidents would not sever proximate causation and absolve a carmaker of liability under OPLA. See Fed. Steel & Wire Corp. , 543 N.E.2d at 775.

OPLA affects the proximate cause analysis for an additional reason. The statute requires manufacturers to foresee and guard against risks that ordinary people may overlook. Ohio Rev. Code Ann. § 2307.75(B)(5) (considering whether product's design is 'more dangerous than a reasonably prudent consumer would expect'); id. § 2307.75(E) (excluding certain products whose inherent danger is 'recognized by the ordinary person with the ordinary knowledge common to the community”); see also id. § 2307.76(B) (excluding products whose risk is “open and obvious” or a “matter of common knowledge”). Assessing foreseeability from an ordinary person’s perspective at the proximate cause stage, therefore, would effectively nullify that core duty. See United States v. Tohono O’Odham Nation , 563 U.S. 307, 315 (2011) (“Courts should not render statutes nugatory through construction.”).

In short, lack of foreseeability is the 'underlying rationale' for Ross . Pendrey , 479 N.E.2d at 283 (discussing Ross , 203 N.E.2d at 120). But in assessing foreseeability, Ross and the cases applying it view risk from the vantage point of those who own cars. They do not address what a manufacturer designing cars knows or should know when it comes to the risks of theft-related accidents. Yet the defendants here, Kia and Hyundai, are sophisticated car manufacturers. And under Ohio law, we must assess foreseeability from the perspective of an actor in their positions.

The difference in the perspective of a car manufacturer and a car owner (or even a car dealership) matters. For example, in Pendrey , the Ohio Supreme Court declined to hold a car owner liable based on government statistics on theft-related accidents. See 479 N.E.2d at 284. The statistics concerned accident rates for thousands of stolen cars across the country, not any single car in a particular locality. Nothing in Pendrey , moreover, suggested that the car owner knew or should have known about those statistics. Nor would the law otherwise impute knowledge of those statistics to the average person. See Tompkin v. Am. Brands , 219 F.3d 566, 569, 572–73, 575 (6th Cir. 2000). But the same considerations do not apply to manufacturers. A manufacturer “ha[s] an obligation to be an expert in its product, which includes the testing Page 9and monitoring of known and possible hazards.” Boyd v. Lincoln Elec. Co. , 902 N.E.2d 1023, 1028 (Ohio Ct. App. 2008). So, in product liability cases, plaintiffs can point to general industry knowledge, research literature, and statistical evidence to show the risks that a manufacturer, as an expert, knew or reasonably should have known. See Krumpelbeck v. Breg, Inc. , 491 F. App’x 713, 719 (6th Cir. 2012); 1 CV Ohio Jury Instructions 451.07 (manufacturer’s awareness of risk depends on “state of medical, scientific, and technical research and knowledge”).

Strench and Moshi draw on similar industry-specific materials to support foreseeability in these cases. Most significantly, Strench and Moshi point to the federal standard that governs anti-theft designs in passenger vehicles. That standard explicitly aims to “reduce the incidence of crashes resulting from theft.” 49 C.F.R. § 571.114. It was adopted more than 50 years ago in response to the disproportionate rate of accidents among stolen vehicles. See 33 Fed. Reg. 6471, 6471 (Apr. 27, 1968). In the preamble to the rule, the National Highway Traffic Safety Administration cited evidence that '18.2 percent of . . . stolen cars became involved in accidents.” Id . It thus concluded that stolen vehicles created “a major hazard to life and limb on the highways,' and that reducing car theft would 'protect the many innocent members of the public who are killed and injured by stolen cars each year.” Id. Based on these allegations, a jury could conclude that Kia and Hyundai, as regulated and sophisticated entities, knew or reasonably should have known about the connection between car theft and accidents. See Covenant Med. Ctr., Inc. v. Sebelius , 424 F. App’x 434, 439 (6th Cir. 2011).

Strench and Moshi also argue that the Kia Challenge itself put Kia and Hyundai on notice of the risk of theft-related accidents. Kia and Hyundai disagree, suggesting that the “unprecedented social-media theft craze” makes the injuries here unforeseeable. Kia Br. at 19; Hyundai Br. at 19. Neither side is entirely correct. The Kia Challenge does not bear on foreseeability for the design defect claims because it emerged years after the companies made the stolen cars in these cases. But it also does not negate foreseeability because the surge in thefts began even before the Kia Challenge, and in any event, Kia and Hyundai did not have to foresee the precise way in which theft-related accidents would come about to be liable. Mudrich v. Standard Oil Co. , 90 N.E.2d 859, 864 (Ohio 1950); Oiler v. Willke , 642 N.E.2d 667,Page 10 672 (Ohio Ct. App. 1994). In short, the Kia Challenge does not move the foreseeability analysis in either side’s direction.

Our conclusion that a manufacturer can be liable despite a third party's intervening misconduct does not break new ground. Ohio courts have come to the same conclusion in other cases. For example, a gun manufacturer may violate OPLA by failing to “incorporate feasible safety devices that would prevent unauthorized use' by third parties. Beretta , 768 N.E.2d at 1145-46. An elevator manufacturer may be liable for third-party injuries even if a 'vandal' had tampered with its elevator. See R.H. Macy & Co. v. Otis Elevator Co. , 554 N.E.2d 1313, 1316–17 (Ohio 1990); Volter v. C. Schmidt Co. , 598 N.E.2d 35, 38-39 (Ohio Ct. App. 1991). And a sophisticated oil company may be liable for its customers' negligent driving on its property. Hollingsworth v. Standard Oil Co. of Ohio , No. 8815, 1978 WL 215387, at *3 (Ohio Ct. App. Sept. 27, 1978) (noting the business's greater 'experience, planning, and technical expertise” than a “regular customer”). What matters is not that the claim arises from third-party misconduct, but that the manufacturer should have foreseen the misconduct given its knowledge of its product’s risks.

Nor do we depart from Ohio’s “key-in-the-car” precedents in holding that a defendant’s knowledge matters. The dealership cases that Kia and Hyundai rely on show why. In one case, the plaintiffs sued a dealership based only on the dealership’s failure to secure its cars against theft. Tilton , 1997 WL 816526, at *1. The court rejected the claim because, under Pendrey , the mere fact that the dealership failed to secure its cars spoke to the foreseeability of theft, but not theft-related accidents. Id. at *3. Compare that with another case, where the dealership left a disgruntled employee, who was drunk and erratic, 'in a lot full of cars with the keys in the ignition.' Fletcher Trucking , 1995 WL 360323, at *5. The court declined to apply Pendrey because the dealership “knew that [its employee] was intoxicated,” and that knowledge made not just theft by the employee but also an accident foreseeable. Id. at *6.³

Page 11Viewed together, Ohio’s “key-in-the-car” cases hold that a car owner’s failure to secure a car against theft, without additional allegations of specific knowledge, does not create liability for accident-related injuries. But the cases do not bar liability where additional evidence puts a defendant, including a sophisticated manufacturer, on notice of the likelihood of theft-related accidents. See id. ; Inskeep v. Columbus Zoological Park Ass’n , 207 N.E.3d 876, 883–84 (Ohio Ct. App. 2023) (declining to apply Pendrey where zoo employee left golf cart unattended around children and toddler drove it and injured another); cf. Fed. Steel & Wire Corp. , 543 N.E.2d at 775 (approvingly citing case that held company liable “for injuries caused by the unauthorized use of its bulldozer” because “past incidents” had put company on notice of risk).

Finally, the carmakers argue that Ross and Pendrey reflect a broader 'policy judgment' about who should bear responsibility in cases like these. Kia Br. at 1; Hyundai Br. at 1. Ross and Pendrey declined to make car owners “insurer[s]” for a thief’s tortious conduct, leaving that task to the legislature instead. Ross , 203 N.E.2d at 121; Pendrey , 479 N.E.2d at 284. The carmakers urge us to limit OPLA in the same way. They warn that our failure to do so would create what they view as a “novel” change in state law. Kia Br. at 9; Hyundai Br. at 9–10.

Yet it is the carmakers’ argument that would expand Ohio law in new directions. Ross and Pendrey were simple negligence cases involving a single car. These lawsuits raise product liability claims based on the marketing of millions of allegedly defective cars. The carmakers do not explain why we should treat these cases alike as a policy matter. See McFarland v. Bruno Mach. Corp. , 626 N.E.2d 659, 660 & n.2, 664 (Ohio 1994) (declining to extend policy-based rules from negligence law to OPLA). And we could not do so even if we agreed with the carmakers. When Ohio enacted OPLA, the statute became the definitive statement of its policy in this area. See Stiner v. Amazon.com, Inc. , 164 N.E.3d 394, 401 (Ohio 2020). That means in OPLA cases, we cannot default to common law policies and wait for legislative correction. We must discern the state’s policy “from the text of the Act itself” and turn to common law policies only if the text invites it. Id. Relevant here, OPLA says nothing about limiting manufacturers’ liability based on policies from negligence law. But it does say that manufacturers can face Page 12liability for harms that ordinary people do not have to guard against. Importing policy-based limits on ordinary car owners' liability into OPLA would override that state judgment, not respect it.

Even if Ross and Pendrey 's policy concerns applied to OPLA, they would not counsel against proximate causation here. At bottom, Ross and Pendrey rejected a rule that would hold car owners strictly liable based on “mere ownership” of a car. Ross , 203 N.E.2d at 121; see also Pendrey , 479 N.E.2d at 284. But liability that turns on a defendant’s particular knowledge is not strict liability. See McFarland , 626 N.E.2d at 661. And the fact that manufacturers often know about their product’s risks does not make it so.

In sum, at this initial stage, we think Strench and Moshi have alleged enough facts to make foreseeability plausible in these cases. We thus hold that their claims do not fail as a matter of law for lack of proximate causation.⁴

Apart from generally challenging foreseeability, Kia and Hyundai do not present any other argument against Strench’s and Moshi’s design defect claims. They do argue, however, that Strench and Moshi otherwise fail under OPLA to plausibly allege the required elements for their manufacturing, nonconformance, and warning defect claims. The district court did not reach these claim-specific arguments. But we can affirm on any ground supported by the record, so we address each argument in turn. Does v. Whitmer , 69 F.4th 300, 305 (6th Cir. 2023).

Strench and Moshi allege that the cars at issue suffered from manufacturing defects. A manufacturing defect, as its name suggests, is an error in the production process. See Ohio Rev. Code Ann. § 2307.74. It exists if, at the time a product left the control of its manufacturer, the product departed from the manufacturer's intended result or differed from otherwise identical units. See id. ; Donegal Mut. Ins. v. White Consol. Indus., Inc. , 852 N.E.2d 215, 226 (Ohio Ct. App. 2006).

Strench and Moshi do not allege any facts that suggest the cars here deviated in any relevant way from their intended design or identical units. Quite the opposite. They allege that Kia and Hyundai 'knowingly manufactured' entire lines of vehicles that uniformly lacked standard anti-theft features. Moshi Compl., R. 17, PageID 73-74; Strench Compl., R. 19, PageID 57. Those allegations speak to a claim 'for a design defect rather than for a manufacturing defect.” Atkins v. Gen. Motors Corp. , 725 N.E.2d 727, 731 (Ohio Ct. App. 1999). We therefore affirm the dismissal of Strench’s and Moshi’s manufacturing defect claims.

Strench and Moshi also assert claims for nonconformance to representation. See Ohio Rev. Code Ann. § 2307.77. Under OPLA, “[a] product is defective if it did not conform, when it left the control of its manufacturer, to a representation made by that manufacturer.” Id. Apart from proximate causation, nonconformance claims require (1) an express and material representation by the manufacturer about its product's character, quality, or safety; (2) the product's nonconformance to that representation; and (3) the plaintiff's justifiable reliance on that representation. See Gawloski v. Miller Brewing Co. , 644 N.E.2d 731, 734 (Ohio Ct. App. 1994); Ohio Rev. Code Ann. § 2307.71(A)(14).

Strench and Moshi have identified no express representation by Kia and Hyundai about their cars’ safety—the very first element of their nonconformance claims. Krumpelbeck , 491 F. App’x at 722. They merely allege that Kia and Hyundai “represent[ed] their cars are safe, and as safe or safer than cars made [by] other manufacturers.” Moshi Compl., R. 17, PageID 80; Strench Compl., R. 19, PageID 64. But that allegation does little more than restate the element.

Page 14Without “further factual enhancement,” it is too conclusory to state a claim. Iqbal , 556 U.S. at 678 (citation omitted). Strench’s and Moshi’s nonconformance claims thus fail.

Finally, Strench and Moshi allege that Kia and Hyundai failed to adequately warn purchasers about their cars' alleged anti-theft defect or instruct consumers on how to prevent their cars from being stolen. OPLA requires manufacturers to give adequate 'warning or instruction' regarding the risks associated with their products. See Ohio Rev. Code Ann. § 2307.76. Kia and Hyundai contend that the complaints do not allege 'concrete facts' to support the idea that a warning or instruction would have affected consumer behavior. Kia Br. at 27–28; Hyundai Br. at 28. For that reason, they argue, the claims fail on proximate causation.

When it comes to warning claims, Ohio law presumes that an inadequate warning or instruction is the proximate cause of a plaintiff’s injury. Boyd , 902 N.E.2d at 1032 (citing Seley v. G. D. Searle & Co. , 423 N.E.2d 831, 838 (Ohio 1981)); Miles v. Kohli & Kaliher Assocs., Ltd. , 917 F.2d 235, 249 (6th Cir. 1990). The manufacturer has the burden of rebutting that presumption through evidence that a consumer would have disregarded a warning or instruction if it had been provided. Miles , 917 F.2d at 249. At this stage, we must limit our review to the allegations in the complaints, which, unsurprisingly, present no fact that could rebut the applicable presumption. See Bassett v. Nat'l Collegiate Athletic Ass'n , 528 F.3d 426, 430 (6 th Cir. 2008). Accordingly, we reject Kia and Hyundai's argument and reverse the district court’s dismissal of these claims.

We affirm the district court’s dismissal of Strench’s and Moshi’s manufacturing defect and nonconformance claims. We reverse the dismissal of their design defect and warning claims and remand for further proceedings consistent with this opinion.

Page 15MURPHY, Circuit Judge, concurring in the judgment in part and dissenting in part. Two well-established legal rules-one state, the other federal-show that we should affirm the judgments in these diversity cases. As for a summary: Ohio courts have repeatedly held that defendants who negligently facilitate a car theft are not the proximate cause of any accident that the thief causes when recklessly driving the stolen car. See Pendrey v. Barnes , 479 N.E.2d 283, 283-84 (Ohio 1985) (per curiam); Ross v. Nutt , 203 N.E.2d 118, 120-21 (Ohio 1964). And federal courts have just as repeatedly refused to adopt a new interpretation of state law that expands a defendant's liability when the state courts have yet to make the change. See US Framing Int'l LLC v. Cont'l Bldg. Co. , 134 F.4th 423, 435 (6th Cir. 2025). In short, the plaintiffs in these cases should have sued in state court because they seek novel changes to Ohio proximate-causation law. Since my colleagues nevertheless allow some of their claims to proceed, I respectfully dissent in part.

These two cases have the same tragic facts. Over several years, Hyundai Motor America and Kia America failed to put “engine immobilizers” (common anti-theft devices) in many cars. Immobilizers have curbed the rates of car theft at a modest cost. So the lack of immobilizers made Hyundais and Kias 'prime' theft targets. Moshi Compl., R.17, PageID 75; Strench Compl., R.19, PageID 59. In 2022, “young teenage boys” also began to post “viral videos” on social media providing step-by-step instructions on how to steal these cars. Moshi Compl., R.17, PageID 76; Strench Compl., R.19, PageID 60. The wide dissemination of these videos created a spike in Hyundai and Kia thefts by teenagers who came to be known as 'Kia Boys.' Moshi Compl., R.17, PageID 76; Strench Compl., R.19, PageID 60.

Yet victims of theft do not bring these cases against Hyundai and Kia. The cases instead arise from accidents that the thieves caused while recklessly driving the stolen cars. In March 2023, an unknown teen stole a 2019 Hyundai Elantra. Strench Compl., R.19, PageID 61. ThePage 16 teen recklessly drove the Hyundai into Donald Strench’s car and seriously injured Strench. Id. , PageID 61, 65. Similarly, in November 2023, a “Kia Boy” stole a 2018 Kia Optima. Moshi Compl., R.17, PageID 77. The thief recklessly crashed the Kia into Matthew Moshi’s car, killing Moshi. Id. Strench and Moshi's estate sued Hyundai and Kia (respectively), alleging that the failure to include immobilizers in their cars violated (as relevant now) the Ohio Product Liability Act. A district court dismissed their complaints on the ground that Hyundai’s and Kia’s failure to use this technology was not the proximate cause of the harms that Strench and Moshi suffered. See Moshi v. Kia Am., Inc. , 2024 WL 3178689, at *2-3 (S.D. Ohio. June 26, 2024); Strench v. Hyundai Motor Am. Corp. Serv. Co. , 2024 WL 3178681, at *2–3 (S.D. Ohio June 26, 2024).

I agree with the district court that these claims should fail under settled Ohio law. The parties do not dispute that the Ohio Product Liability Act follows the “proximate causation” rules from Ohio's common law. See Ohio Rev. Code § 2307.73(A)(2); R.H. Macy & Co. v. Otis Elevator Co. , 554 N.E.2d 1313, 1316 (Ohio 1990). As a generic matter, the Ohio Supreme Court has held that a defendant’s negligence qualifies as a proximate cause of a plaintiff’s injury if the injury was the “ordinary and natural result[]” of the negligence, such that the defendant should have reasonably “expected” (or foreseen) it. Neff Lumber Co. v. First Nat’l Bank , 171 N.E. 327, 329 (Ohio 1930) (citation omitted); see R.H. Macy , 554 N.E.2d at 1316; Mussivand v. David , 544 N.E.2d 265, 272 (Ohio 1989); Ross , 203 N.E.2d at 120. The court has added that a defendant's negligence sometimes will not meet this test if a third party's intervening actions stand in between the negligence and injury. See Queen City Terminals, Inc. v. Gen. Am. Transp. Corp. , 653 N.E.2d 661, 670 (Ohio 1995); Pendrey , 479 N.E.2d at 283-84; R.H. Macy , 554 N.E.2d at 1316–17.

As best I can tell, the Ohio Supreme Court has used different intervening-cause tests for different intervening causes. In some cases, the court has framed the test solely in foreseeability terms. See Jeffers v. Olexo , 539 N.E.2d 614, 618 (Ohio 1989); Strother v. Hutchinson , 423 N.E.2d 467, 470-71 (Ohio 1981) (per curiam). These cases hold that a defendant's negligence is not a proximate cause unless the defendant 'knew or, acting as a reasonably prudent person, should have known” that a third party might take the final step that leads to the Page 17plaintiff’s injury. Jeffers , 539 N.E.2d at 618. In other cases, the court has looked to the third party’s (not the defendant’s) mindset. See Mussivand , 544 N.E.2d at 273; Cascone v. Herb Kay Co. , 451 N.E.2d 815, 819 (Ohio 1983). These cases hold that a defendant’s negligence is not a proximate cause if a “conscious and responsible” third party “could or should have eliminated the hazard' created by that negligence before the plaintiff suffered an injury. Mussivand , 544 N.E.2d at 273 (quoting Cascone , 451 N.E.2d at 819). In a third set of cases, the court has articulated a distinct two-part test. See Queen City Terminals , 653 N.E.2d at 671; Leibreich v. A.J. Refrigeration, Inc. , 617 N.E.2d 1068, 1071-72 (Ohio 1993). These cases hold that a defendant’s negligence is not a proximate cause if a third party takes a “ new ” and “ independent ” action. Queen City Terminals , 653 N.E.2d at 671 (quoting Cascone , 451 N.E.2d at 819). The cases find a third party’s action “independent” of a defendant’s negligence if no “relationship of cause and effect” exists between the two. Id. (quoting Leibreich , 617 N.E.2d at 1072). And they find a third party’s actions “new” if the defendant could not have “reasonably” “foreseen” them. Id.

It is unclear to me why the Ohio Supreme Court chooses one intervening-cause test in some cases and another in other cases. Until we receive greater guidance from that court, then, I would stick closely to the precedent that has addressed an intervening cause most like the intervening cause in the case under review. And there can be no doubt which decisions consider an intervening cause most like the one in these car-accident cases: Ross and Pendrey .

In those cases, the Ohio Supreme Court held that a defendant whose negligence helps a thief steal a car cannot face liability for injuries caused by the thief's reckless driving. See Pendrey , 479 N.E.2d at 283–84; Ross , 203 N.E.2d at 120–21. Car owners in the cases had left their unattended cars with the key in the ignition-in violation of a state law or city ordinance. See Pendrey , 479 N.E.2d at 283–84; Ross , 203 N.E.2d at 119. Thieves then stole the cars, drove them recklessly, and got into accidents that harmed others. See Pendrey , 479 N.E.2d at 283–84; Ross , 203 N.E.2d at 119. The injured parties sued the car owners for their negligence in allowing the thieves to recklessly drive the cars. See Pendrey , 479 N.E.2d at 283–84; Ross , 203 N.E.2d at 119.

Page 18The Ohio Supreme Court decided Ross first. There, it set the legal rule that a car owner’s negligent failure to remove the key from the ignition of an unattended car is not a proximate cause of another person’s injuries when the thief drives the car recklessly. See Ross , 203 N.E.2d at 120-21. The court assumed that the owner had acted negligently, so the owner's liability rested on causation alone. See id. at 120. Which proximate-cause test did the court apply? At first blush, it appeared to ask a foreseeability question: Could the car owner 'reasonably have foreseen” that his negligence would lead the thief to steal the car and recklessly drive it? Id. But its reasoning did not follow this path. In particular, the court did not ask about the specific owner's state of mind-considering, for example, whether he knew or should have known anything about the accident rates of thieves. Rather, it categorically held that a thief's negligent driving was “an efficient intervening cause” that broke “the chain of causation.” Id. The court reasoned that the car owner's failure to remove the key from the ignition was 'merely a condition” that the thief exploited to cause the harm and that the owner was “not in law bound to foresee' the thief's negligent driving. Id. at 121 (quoting Clements v. Tashjoin , 168 A.2d 472, 475 (R.I. 1961)) (emphasis added). Why? It supported this conclusion with a policy argument—not a foreseeability argument. The court refused to make a car owner “an insurer of the public” for harms that the car causes when driven “by a third person” without the owner’s permission. Id. at 121. By doing so, it followed the majority rule across the States. Id. (citing cases); see also Cornelius J. Peck, An Exercise Based Upon Empirical Data: Liability for Harm Caused by Stolen Automobiles , 1969 Wis. L. Rev. 909, 910.

The Ohio Supreme Court reaffirmed this holding over 20 years later in Pendrey . See 479 N.E.2d at 284. There, the court considered two legal developments and one factual development since Ross . Legally, the Ohio Supreme Court had issued a new precedent ( Strother ), and the Ohio legislature had issued a new law (Ohio Revised Code § 4511.661(A)). In Strother , the court held that a defendant who shot a robber fleeing in a car could face tort liability to those that the robber crashed into because of the shooting. 423 N.E.2d at 471. In Ohio Revised Code § 4511.661(A), the Ohio legislature made it a crime for drivers to leave an unattended car with the key in the ignition. Yet neither change led Pendrey to depart from Ross . Pendrey reasoned that the Strother defendant's shooting had caused the robber's crash, but the car owner’s failure to remove the keys had not “induced negligent or reckless behavior on the Page 19part of the car thief once the car had already been taken.” 479 N.E.2d at 284. And Pendrey did not interpret the new Ohio law as designed to protect against the harms caused by a thief's negligent driving. Id.

Factually, the U.S. Department of Justice had gathered empirical information showing that thieves driving stolen cars are far more likely to get into accidents than other drivers. See id. ; see also Peck, supra , at 915. The plaintiff in Pendrey relied on these statistics to argue that a reasonable jury could now find that the car owner there could have foreseen the thief’s accident. 479 N.E.2d at 284. But the court made short work of this argument. It did not disagree with the foreseeability premise. It instead reiterated Ross ’s policy argument, refusing to rely on “judicial fiat” to make every car owner an “insurer” for the harms that car thieves cause. Id. And it found this empirical evidence “better directed towards the legislature than the courts.” Id.

Ross and Pendrey now represent settled Ohio law. Over the years, these two decisions have foreclosed many tort suits that rested on similar fact patterns. See Lagowski v. Shelly & Sands, Inc. , 38 N.E.3d 456, 458–61 (Ohio Ct. App. 2015); Bugaj v. Nationwide Ins. , 2009 WL 2678938, at *1–2 (Ohio Ct. App. Aug. 28, 2009); Magill v. Moore , 2006 WL 1843627, at *2–3 (Ohio Ct. App. July 3, 2006); Tilton v. Austin Motors, Inc. , 1997 WL 816526, at *2–4 (Ohio Ct. App. Dec. 30, 1997); Brubaker v. Grzechowiak , 1986 WL 12232, at *1, *4-5 (Ohio Ct. App. Oct. 31, 1986); Burchett v. Model Dairy of Chillicothe , 1982 WL 3541, at *1 (Ohio Ct. App. Oct. 4, 1982); Fonticella v. Davis , 1980 WL 355345, at *1-2 (Ohio Ct. App. Nov. 6, 1980); Miles v. Cedar-Lee Chrysler-Plymouth, Inc. , 1978 WL 218347, at *1–2 (Ohio Ct. App. April 27, 1978); Watson v. Bobb Chevrolet Co. , 1974 WL 184174, at *2 (Ohio Ct. App. June 18, 1974).

The decisions should foreclose these two diversity suits too. The suits involve the same intervening actors as those in Ross and Pendrey . Third parties first committed an intentional crime (the theft of a car) and then engaged in negligent conduct (reckless driving) that harmed others on the road. And like the plaintiffs in Ross and Pendrey , the victims in these cases do not seek to hold the defendants liable for the thefts ; they seek to hold them liable for the negligent driving . Yet the lack of an engine mobilizer-like a key in the ignition-was 'merely a condition” that the thief used to commit the “intervening independent negligence” that caused the accident; it was not the type of “dangerous condition” that can itself qualify as a cause of that Page 20accident. Ross , 203 N.E.2d at 121 (quoting Clements , 168 A.2d at 475); see Richard A. Epstein, A Theory of Strict Liability , 2 J. Legal Stud. 151, 179 (1973) (noting tort distinction between “dangerous” and “mere” condition). Likewise, a decision not to put an engine immobilizer in the car (like a decision to leave the key in its ignition) did not 'induce[] negligent or reckless behavior on the part of the car thief once the car had already been taken.” Pendrey , 479 N.E.2d at 284. Put differently, the defendants have a “duty to guard against” only the risk that makes their actions negligent. H.L.A. Hart & Tony Honoré, Causation in the Law 203 (2d ed. 1985). And unlike some other state courts, the Ohio Supreme Court has defined this risk narrowly to include only the risk of theft—not the risk of negligent driving after the theft. Id. ; see Pendrey , 479 N.E.2d at 284. The court did so because it refused to make a car owner “an insurer of the public” for harms caused by those who drive their cars. Ross , 203 N.E.2d at 121. The same logic covers these cases.

My colleagues distinguish Ross and Pendrey on the sole ground that the decisions involved car owners , while these cases involve car manufacturers . Even if a reasonable owner might not foresee that a thief could negligently drive a stolen car, their logic goes, a jury could still find that a reasonable manufacturer knew or should have known of this risk. I disagree with this distinction. To start, one will search Ross and Pendrey in vain for any suggestion that the defendant’s status mattered. The decisions instead adopted a categorical rule: a thief’s negligent driving is “an efficient intervening cause” that breaks the chain of causation between the accident and earlier acts of negligence. Ross , 203 N.E.2d at 120. Under my colleagues' reading, by contrast, the plaintiffs in Ross and Pendrey simply sued the wrong defendants. Rather than suing the owners, they should have sued the manufacturers for negligently designing the cars to allow the owners to exit with the key in the ignition (at least without a warning sound or the like). Cf. Peck, supra , at 925; 33 Fed. Reg. 6471, 6471 (Apr. 27, 1968). Yet how could the thief break the causal chain for an actor closer to the injury’s source (the car owner) but not one farther away from it (the car manufacturer)?

Unlike other Ohio Supreme Court cases, Ross and Pendrey also did not rely on the 'knew or should have known” test that my colleagues invoke. See Jeffers , 539 N.E.2d at 618. This test would require Ohio courts to undertake a case-by-case analysis that looks to each specific Page 21defendant’s knowledge. Cf. Queen City Terminals , 653 N.E.2d at 671; Leibreich , 617 N.E.2d at 1072. But the Ohio Supreme Court held categorically that no owners can face liability. See Pendrey , 479 N.E.2d at 284. So even if a car manufacturer’s CEO negligently leaves his key in the ignition with full awareness of the risk that a thief will drive the car recklessly, the CEO would not count as the proximate cause of any accident that results from that driving.

Pendrey confirms my point in two ways. After noting that the defendant in Strother “knew, or should have known,” that his shooting of a fleeing robber might cause the robber to crash into others, the court did not then ask whether every car owner categorically lacked knowledge that thieves might drive recklessly—an implausible assertion. 479 N.E.2d at 284. It instead shifted the legal focus. The court distinguished Strother because the negligent conduct of leaving a key in the ignition (unlike the negligent conduct of shooting a driver in a moving car) did not itself cause the thief to drive recklessly. Id. The same categorical rule applies to the lack of engine immobilizers: this alleged defect likewise did not “induce[]”reckless driving. Id.

Next, when confronted with statistics showing that thieves get into far more car accidents than other drivers, Pendrey did not respond, as my colleagues do, that car owners would not have known of this empirical information. (Frankly, I think most ordinary individuals who have watched movies with high-speed chases of stolen cars would agree about the greater risks involved.) The court instead made what was 'essentially a policy determination' that Ohio courts should not impose liability on individuals whose negligence facilitates a car theft for the injuries that the thief later causes. Hunt v. Marksman Prods. , 656 N.E.2d 726, 728 (Ohio Ct. App. 1995). Pendrey was unwilling to make car owners 'insurer[s]' of criminals who intentionally capitalize on their negligence. 479 N.E.2d at 284. And its response to the statistics showing that thieves get into more accidents applies in full force here: plaintiffs should direct this information to Ohio’s “legislature”—not the federal “courts”—for any change in law. Id. By relying on foreseeability arguments that played no part in Pendrey , my colleagues now make every car manufacturer the type of criminal insurer that the state court found 'unacceptable.' Id. I also wonder about the limits of my colleagues’ foreseeability arguments. Thieves often use stolen cars to commit other crimes—such as bank robberies. If empirical evidence backed up Page 22this instinct, should a jury get to decide whether to hold Kia and Hyundai liable to the robbed banks for the heists?

The other Ohio cases applying Ross and Pendrey confirm my reading of those decisions. Ohio courts have extended the decisions not just to ordinary car owners but also to more sophisticated entities that likely know thieves often recklessly drive stolen cars. To give a few examples, a construction company was not the proximate cause of a plaintiff’s bulldozed house when a thief stole an unsecured bulldozer from a construction site and destroyed the house. See Lagowski , 38 N.E.3d at 458–61. A car dealer was not the proximate cause of an accident when a thief stole an unsecured car from the dealer’s lot and drove it into the plaintiff’s vehicle. See Tilton , 1997 WL 816526, at *2–4; see also Miles , 1978 WL 218347, at *1–2; Watson , 1974 WL 184174, at *2. And a dairy company was not the proximate cause of an accident when a thief stole an unattended milk truck and ran it into the plaintiffs. Burchett , 1982 WL 3541, at *1. In none of these cases did the court care about the defendant’s sophistication level.

My colleagues distinguish this precedent by relying on Ohio cases applying a “special duty” test. Stubbs v. Hodge , 1996 WL 732543, at *3 (Ohio Ct. App. Dec. 19, 1996); see Inskeep v. Columbus Zoological Park Ass'n , 207 N.E.3d 876, 883 (Ohio Ct. App. 2023); Fletcher Trucking v. Columbus Fair Auto Auction, Inc. , 1995 WL 360323, at *6 (Ohio Ct. App. June 13, 1995). In these cases, the defendant (say, a business) has a special relationship to a tortfeasor (say, an employee) or a tortfeasor's victim (say, a business invitee). See Stubbs , 1996 WL 732543, at *3. This special duty requires the defendant to “control” the tortfeasor to prevent harm to the victim. Id. The duty got triggered, for example, when an employer took 'control' of an employee who had become intoxicated during the workday. Fletcher , 1995 WL 360323, at *4–5. An Ohio court held that the employer could face liability when the employee stole a car from its premises and got into an accident with the victim. Id. at *5-6. It makes sense for Ross and Pendrey not to bar suits in this special-duty scenario because the employer must protect against any harm from the employee-whether caused by the employee's driving, the employee's assault, or any other misconduct. But the special-duty logic does not apply here because car manufacturers do not have any special 'relationship' with car thieves or their victims. Id. at *5. Do my colleagues hold instead that manufacturers have a duty to 'control' all Page 23unknown thieves simply because they make cars? Stubbs , 1996 WL 732543, at *3. That is quite an expansion of this special-duty caselaw.

Their reliance on the Ohio Product Liability Act (and cases interpreting it) fares no better. They point out that whether a product 'design' is 'defective' accounts for the 'foreseeable risks” of the design evaluated from the perspective of a “reasonable manufacturer[.]” Ohio Rev. Code §§ 2307.75(A); 2307.71(A)(6)(b)(i); see also Krumpelbeck v. Breg, Inc. , 491 F. App'x 713, 719 (6th Cir. 2012); Boyd v. Lincoln Elec. Co. , 902 N.E.2d 1023, 1028 (Ohio Ct. App. 2008). This reasoning mistakes an irrelevant question (did the lack of engine immobilizers qualify as a design defect ?) for the relevant one (does that defect qualify as the proximate cause of the harms?). Indeed, whether Kia and Hyundai manufactured cars with a design defect (because they failed to use immobilizers) is analogous to whether the car owners in Pendrey and Ross acted negligently (because they left the key in the ignition). But those cases merely “assum[ed] negligence on the part of the” car owners when finding them not liable on causation grounds. Ross , 203 N.E.2d at 120. I would do the same by assuming that the cars here had a defect-which does nothing to answer the separate causation question. If anything, my colleagues’ logic violates basic rules of statutory interpretation. The Ohio Product Liability Act permits a plaintiff to obtain damages only if the plaintiff proves both that the product 'was defective in design” and that the defect 'was a proximate cause' of the injury. Ohio Rev. Code § 2307.73(A)(1)–(2). These are separate elements. By relying on the defect element to find the cause element met, they give the latter element no “independent work” to do. Pulsifer v. United States , 601 U.S. 124, 142 (2024).

The other Ohio Supreme Court decisions that my colleagues cite also do not change things. See Cincinnati v. Beretta U.S.A. Corp. , 768 N.E.2d 1136, 1145-49 (Ohio 2002); R.H. Macy , 554 N.E.2d at 1316-17. In Beretta , the Ohio Supreme Court allowed a city's complaint to proceed against gun manufacturers when it alleged that the manufacturers had engaged in practices that “fostered the criminal misuse of firearms” and thereby increased the city's expenses for public services. 768 N.E.2d at 1140. In the court's causation analysis, though, it considered only whether the city’s injuries were too remote—not whether any third parties should qualify as superseding causes-as shown by its failure to cite (let alone Page 24distinguish) Ross and Pendrey . See id. at 1147-49. Besides, I find Beretta a poor predictor of Ohio law because the Ohio legislature has all but overruled the decision by preempting the common-law claims at issue in that case. See In re Nat’l Prescription Opiate Litig. , __ N.E.3d __, 2024 WL 5049302, at *3 (Ohio Dec. 10, 2024).

As for R.H. Macy , the Ohio Supreme Court there upheld the reversal of a trial judgment against a product manufacturer because the jury instructions failed to include an interveningcause instruction. 554 N.E.2d at 1316-17. In the process, the court seemed to approve the distinct two-part intervening-cause test (that a third party must commit a new and independent act) that some of the court’s other cases have followed. Id. But Ross and Pendrey did not use that test, and I would follow the analysis in those cases because they considered an identical intervening cause.

At the least, federal law makes clear that our court should not extend Ohio proximate causation in the way that Strench and Moshi's estate request. Plaintiffs who choose to bring state-law claims in federal courts must recognize that those courts are 'extremely cautious' about making a “substantive innovation” to state law. Combs v. Int’l Ins. Co. , 354 F.3d 568, 578 (6th Cir. 2004) (citation omitted). We have thus repeatedly held that we must choose an interpretation of state law that “reasonably restricts” a defendant’s “liability” over an interpretation that “greatly expands” it when put to a choice between the two. In re Darvocet, Darvon, & Propoxyphene Prods. Liab. Litig. , 756 F.3d 917, 937 (6th Cir. 2014) (citation omitted); see US Framing , 134 F.4th at 435; Boyd Cnty. ex rel. Hedrick v. MERSCORP, Inc. , 614 F. App'x 818, 823 (6th Cir. 2015). '[B]oth judicial modesty and federalism' principles undergird this rule because it keeps state courts in the driver’s seat over the scope of state law. US Framing , 134 F.4th at 435.

As a result, even if it were debatable whether state courts might distinguish Ross and Pendrey in similar suits against car manufacturers, this federal principle should still foreclose the claims here. In fact, the plaintiffs themselves concede that they seek to make significant statelaw innovations. While acknowledging that their theories “might be novel,” they nevertheless Page 25suggest that the “novelty” of their claims does “not count against” them at the pleading stage. Strench Appellant’s Br. 10, 23; Moshi Appellant’s Br. 9, 22. That may be true in state court. But our marching orders are clear in this federal diversity suit: we must pick the interpretation of state law that “reasonably restricts [the] liability” of Kia and Hyundai. Darvocet , 756 F.3d at 937 (citation omitted). And my reading of Ross and Pendrey is at least a reasonable one. My colleagues, by comparison, implement a significant policy change-potentially making the failure to include engine immobilizers in Ohio vehicles actionable (and so illegal). Whether to take this course involves difficult tradeoffs. Cf. Drew Thornley, America’s “Kia Boys”: The Problem, Responses, and Recommendations , 47 Seattle Univ. L. Rev. 35, 43–48 (2024). Ohio’s accountable legislature or courts-not an isolated federal circuit court-should balance the tradeoffs.

Because I would hold that the claims in these cases fail on proximate-causation grounds, I would not reach the other merits issues that my colleagues address in this appeal. I thus respectfully concur in the judgment in part and dissent in part.