Jeffrey Clark v. Louisville-Jefferson Cnty. Metro Gov't

130 F.4th 571

• Court: Court of Appeals for the Sixth Circuit
• Decided: March 7, 2025
• Docket: 24-5061
• Status: Published

Opinion

RECOMMENDED FOR PUBLICATION Pursuant to Sixth Circuit I.O.P. 32.1(b) File Name: 25a0050p.06

UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT

┐ JEFFREY DEWAYNE CLARK, │ Plaintiff, │ │ GARR KEITH HARDIN, │ > No. 24-5061 Plaintiff-Appellee, │ │ v. │ │ │ LOUISVILLE-JEFFERSON COUNTY METRO GOVERNMENT, │ Defendant, │ │ ROBERT THURMAN, Kentucky State Police Crime Lab │ Forensic Serologist, in his individual capacity, │ │ Defendant-Appellant. ┘

Appeal from the United States District Court for the Western District of Kentucky at Louisville. No. 3:17-cv-00419—Gregory N. Stivers, District Judge.

Argued: July 25, 2024

Decided and Filed: March 7, 2025

Before: STRANCH, BUSH, and MURPHY, Circuit Judges. _________________

COUNSEL

ARGUED: Ed Monarch, MCBRAYER PLLC, Louisville, Kentucky, for Appellant. Emma Freudenberger, NEUFELD SCHECK BRUSTIN HOFFMANN & FREUDENBERGER, LLP, New York, New York, for Appellee. ON BRIEF: Ed Monarch, Peter J. Rosene, MCBRAYER PLLC, Louisville, Kentucky, for Appellant. Emma Freudenberger, Anthony Balkissoon, Anna Benvenutti Hoffmann, Mary Katherine McCarthy, NEUFELD SCHECK BRUSTIN HOFFMANN & FREUDENBERGER, LLP, New York, New York, for Appellee.

The court delivered a PER CURIAM opinion. STRANCH, J. (pp. 16–19) and MURPHY, J. (pp. 20–30), delivered separate concurring opinions. No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 2

_________________

OPINION _________________

PER CURIAM. In 1995, a Kentucky jury convicted Garr Keith Hardin and Jeffrey Clark of murdering Rhonda Sue Warford. Robert Thurman, a forensic serologist, testified at their trial that a hair found at the crime scene was “similar” to a sample of Hardin’s hair. After Clark and Hardin spent over two decades in prison, DNA testing proved that this hair was not, in fact, Hardin’s. A state court thus vacated Hardin’s and Clark’s convictions. Clark and Hardin then brought this suit under 42 U.S.C. § 1983 against (among others) Thurman. In discovery, they obtained the “observation notes” that Thurman had written when examining the hairs. These notes suggested that the hair found at the scene might not have matched Hardin’s hair sample in various ways. Hardin claimed that Thurman’s failure to disclose the notes before trial violated his disclosure obligations under Brady v. Maryland, 373 U.S. 83 (1963). The district court denied Thurman’s qualified-immunity defense. On appeal, Thurman argues (1) that his notes were neither exculpatory nor material under Brady, and (2) that the law in the mid-1990s did not clearly establish that Brady’s duty of disclosure applied to scientists. Our precedent deprives us of jurisdiction over Thurman’s first argument. And it also dooms his second argument that Brady did not clearly apply to him. We thus affirm in part and dismiss in part for lack of jurisdiction.

I

On the evening of April 1, 1992, Rhonda Sue Warford visited a Kroger within walking distance of her home in Louisville, Kentucky. See Hardin v. Commonwealth, 396 S.W.3d 909, 910 (Ky. 2013). When she got back to her house, she told her mother that a strange man had harassed her in the parking lot. See id. Around midnight, she left home again and never returned. See id. On April 5, an elderly couple discovered Warford’s body in a field to the southwest of Louisville in Meade County. See id. Warford’s killer had repeatedly stabbed her. Id. No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 3

Warford had been dating Garr Keith Hardin at the time of her death. See id. Hardin claimed that he last saw Warford on March 28 and had last spoken to her over the phone on the afternoon of April 1. Warford’s mother told the police that Warford, Hardin, and Jeffrey Clark, his close friend, had all been “involved in Satanism.” Id. She, for example, suggested that Hardin had used “a needle and thread” to sew a tattoo of an upside-down cross below Warford’s left collarbone. Tr., R.292-12, PageID 4296. The police came to suspect that Hardin and Clark murdered Warford as part of a satanic ritual. Hardin, 396 S.W.3d at 911.

Investigators recovered various hairs from Warford’s body. See id. They sent these hairs (among other evidence) to the Kentucky State Police Crime Lab. Robert Thurman, a forensic serologist at the lab, analyzed the evidence. Thurman compared Hardin’s, Clark’s, and Warford’s hairs with the hairs found on Warford. He issued his final report at the end of April 1992. His report opined that a “Caucasian head hair found on [Warford’s sweatpants] was similar in color and microscopic characteristics to [a sample of Hardin’s hair] and may have common origin.” Rep., R.316-36, PageID 23748.

A week after Thurman issued this report, a Kentucky grand jury indicted Hardin and Clark for Warford’s murder. But the case did not proceed to trial until years later in early 1995. See Hardin, 396 S.W.3d at 911.

At trial, the prosecution identified Hardin and Clark’s shared desire to engage in a “satanic” sacrifice as the “motive” for Warford’s murder. Id. It thus introduced evidence of their satanism, including the “satanic bible and other occult-related objects found at” Hardin’s home. Hardin v. Commonwealth, 2003 WL 21106138, at *3 (Ky. Ct. App. May 16, 2003) (citation omitted); Clark v. O’Dea, 257 F.3d 498, 500–01 (6th Cir. 2001). To support their theory that this satanism had led to Warford’s murder, a detective with the Louisville Metro Police—Mark Handy—testified about a comment that Hardin had allegedly made to him. According to Handy, Hardin said that he had sacrificed animals in the past but that he “got tired of looking at animals and began to want to do human sacrifices.” Commonwealth v. Clark, 528 S.W.3d 342, 346 (Ky. 2017). The State also presented testimony from an inmate named Clifford Capps who had shared a cell with Clark while Clark awaited trial. Capps testified that Clark had No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 4

twice confessed to murdering Warford. Hardin, 396 S.W.3d at 911. The Kentucky Supreme Court described this statement as the “most incriminating” testimony. Id.

The prosecution also introduced two primary pieces of physical evidence. First, the prosecution used the hairs found on Warford’s pants to link Hardin to the crime scene. Thurman testified that “[o]ne of the hairs that [he] found on [Warford’s] sweatpants was similar to the head hair standards from Garr Hardin.” Tr., R.316-48, PageID 23950. He conceded, though, that other hairs found on Warford’s person did not match Hardin, Clark, or Warford. Id. On cross-examination, defense counsel asked Thurman about the allegedly matching hair: “So, you don’t know if it was actually [Hardin’s] hair or not, but it may be?” Tr., R.316-49, PageID 23963. Thurman replied: “Correct.” Id. Second, the prosecution introduced evidence that the police had discovered Warford’s fingerprint in the backseat of Clark’s car. See Hardin, 396 S.W.3d at 911.

In response, Hardin and Clark presented an “alibi defense,” claiming that they had been in Louisville (not Meade County) at the time of the murder. Id. at 911–12. Clark pointed out that Warford had been in his car many times before December 1991, which would explain why the police found her fingerprint there. Id. at 911, 915. Hardin denied making any statement to Detective Handy about human sacrifices. Clark, 528 S.W.3d at 346. He also denied ever engaging in either animal or human sacrifices.

The jury found both men guilty. See Hardin, 2003 WL 21106138, at *1. A series of significant developments unfolded in the decades following the conviction. Soon after trial, Hardin and Clark learned of a letter that Capps had written to another inmate asking this inmate to lie “to bolster Capps’ story” about Clark’s confessions. Hardin, 396 S.W.3d at 912. Hardin and Clark sought a new trial based on this letter, arguing that Capps had perjured himself when claiming that Clark had confessed. See id. The trial court denied their new-trial motion and sentenced both men to life imprisonment. Id. at 910, 912. Although the Kentucky Supreme Court upheld their convictions on direct appeal, it acknowledged that the prosecution had presented a “highly circumstantial” case against them. Id. at 912 (citation omitted). The state and federal courts later rejected Hardin’s and Clark’s various postconviction motions. See Clark, 257 F.3d at 501, 506; Hardin, 2003 WL 21106138, at *1–4. No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 5

Eventually the Innocence Project agreed to represent Hardin and Clark. See Hardin, 396 S.W.3d at 912. This organization asked prosecutors to release various pieces of evidence for DNA testing. See id. In support, it noted that another man (James Whitley) had allegedly confessed to murdering Warford. See id. Prosecutors rejected its request. See id. at 913. The trial court also denied Hardin and Clark’s joint motion to release the evidence for DNA testing. See id. But the Kentucky Supreme Court reversed the trial court’s decision and held that Hardin and Clark were entitled to the testing. Id. at 915.

The test results proved significant. They excluded Hardin and Clark as the source of the hair on Warford’s sweatpants (although they also excluded Whitley). See Clark, 528 S.W.3d at 345. As a result, the trial court vacated Hardin’s and Clark’s convictions. See id at 348. The Kentucky Supreme Court affirmed. Id. The court reasoned that the “hair evidence” was the “only thing” that put Hardin and Clark at the murder scene. Id. at 345. It added that the “match” of the hair on the sweatpants to Hardin had been an “integral” part of the prosecution. Id.

New prosecutors chose not to retry Hardin and Clark after concluding that they now lacked sufficient evidence to convict the men. In 2018, the trial court granted their motion to dismiss the indictments. Authorities released Hardin and Clark after they had served some 22 years in prison.

Once the state trial court vacated their convictions, Hardin and Clark filed this suit in federal court under 42 U.S.C. § 1983. They sued the Louisville-Jefferson County Metro Government, Meade County, and many officers who worked for the police in these jurisdictions. They also sued Thurman, the serologist who worked for the Kentucky State Police.

Litigation in federal and state court revealed that some of these officials may well have committed misconduct. Of most note, Detective Handy (who testified about Hardin’s alleged human-sacrifice remark) pleaded guilty to committing perjury in another murder case. See Handy v. Louisville/Jefferson Cnty. Metro Gov’t, 2024 WL 1146541, at *1 (Ky. Mar. 14, 2024); see also Clark, 528 S.W.3d at 346−47. Louisville and its officials have since settled with Clark No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 6

and Hardin for some $20 million. Claims remain pending against Meade County and its officials for Brady violations and for alleged falsification of evidence and against Thurman.

This appeal involves only the claims against Thurman. We must assess his liability independent of the conduct of the other investigators. See Jane Doe v. Jackson Loc. Sch. Dist. Bd. of Educ., 954 F.3d 925, 934−35 (6th Cir. 2020). Clark’s and Hardin’s claims against Thurman turn on a potential mismatch between what he said in his final report and what he wrote down in observation notes while examining the hairs. The report disclosed that Hardin’s hair sample and the hair on Warford’s sweatpants were “similar in color and microscopic characteristics[.]” Final Rep., R.316-36, PageID 23748. Under the scientific standards at the time, this similarity finding meant that the hairs shared all the same “characteristics,” such as color and density. Thurman Dep., R.316-5, PageID 21119, 21263.

Thurman’s observation notes, by contrast, suggested that the hairs did not match in various ways. The notes described the color of the hair on the sweatpants as “Lt. Br. to Brown” (light brown to brown), but they described the color of Hardin’s sample as just “Brown.” Notes, R.316-8, PageID 21780, 21783. Likewise, the notes described the pigment density of the hair on the sweatpants as “avg.; light” (average or light), but they described the pigment density of Hardin’s sample as “avg. to sl. heavy” (average to slightly heavy). Id. And the notes described the medulla width of the hair on the sweatpants as “med to thick” (medium to thick), but they described the medulla width of Hardin’s sample as “medium.” Id. If these differences accurately described the hairs, Thurman’s supervisor suggested that it would have been “wholly improper and inconsistent with training and practice” to find the hairs similar. Thomas Dep., R.316-9, PageID 21980–81.

Hardin and Clark thus brought several claims against Thurman arising out of this seeming disconnect between his notes and final report. Two claims matter now. The men alleged that Thurman had fabricated evidence by opining at trial that the hairs were similar when, in fact, they were dissimilar. And they alleged that he had violated his Brady duties by failing to turn over the notes before trial. Thurman moved for summary judgment. No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 7

The district court granted summary judgment to Thurman on all claims but one. See Clark v. Louisville-Jefferson Cnty. Metro Gov’t, 2024 WL 55862, at *7, *12 (W.D. Ky. Jan. 4, 2024). It rejected Clark’s claims in their entirety because Thurman’s report had excluded Clark as a source of the hair and so allegedly had not harmed him. See id. at *7, *12. It also rejected Hardin’s fabricated-evidence claim. Id. at *8–9. Thurman had retained an expert who re- compared the hair on Warford’s sweatpants with the sample of Hardin’s hair (as Thurman had done in 1992). See id. at *3. This expert—the only one who compared the hairs—agreed with Thurman that they were similar. Id. at *3, *9. And the court held that Hardin could not show that Thurman gave a “false” opinion unless Hardin produced “expert testimony” to this effect. Id. at *9. The court found that Hardin had, at most, created a fact dispute over whether Thurman’s notes conflicted with his report. Id. But it found that the notes alone would not allow a jury to find that the hairs were, in fact, dissimilar. Id. We currently lack jurisdiction to review these aspects of the court’s decision.

That said, the court reached a different result for Hardin’s Brady claim. Id. at *10−12. All agree that Thurman did not disclose his observation notes because he followed his lab’s official policy to provide only final reports. See id. at *11. But the court held that a reasonable jury could find that Thurman should have “readily” recognized the “exculpatory value” of these notes given their potential inconsistency with his report. See id. The court next suggested that the notes were material to Hardin’s criminal case because Thurman’s testimony evaluated the “sole physical evidence” that could put Hardin at the murder scene. Id. at *12. And because the Supreme Court had clearly established Hardin’s Brady rights before 1992, the district court found that Thurman could not invoke qualified immunity for this constitutional violation. Id. at *11–12.

II

Thurman immediately appealed the denial of his qualified-immunity defense to Hardin’s Brady claim. In Brady, the Supreme Court held that the “prosecution” violates due process if it suppresses evidence that is both “favorable” to a criminal defendant (or § 1983 plaintiff) and “material” to whether the defendant committed the crime. 373 U.S. at 87. To rebut qualified immunity for this type of claim, § 1983 plaintiffs must show two things. They must show that a No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 8

defendant did, in fact, violate Brady. See Lawler ex rel. Lawler v. Hardeman County, 93 F.4th 919, 925 (6th Cir. 2024). And they must show that the existing caselaw “clearly established” this violation at the time the defendant failed to disclose the relevant evidence. See id.

Thurman argues that Hardin has met neither requirement. He initially asserts that Hardin’s Brady claim fails on its constitutional merits. He next asserts that the Brady caselaw at the time of Hardin’s 1995 trial did not clearly establish that a forensic scientist could have a duty to disclose observation notes (even if such a duty exists today). Our caselaw largely deprives us of jurisdiction over Thurman’s first argument. And it requires us to reject his second one.

A

Thurman first argues that he did not commit a Brady violation even under today’s standards. A Brady claim has three well-known elements. See Jackson v. City of Cleveland, 925 F.3d 793, 814 (6th Cir. 2019). First, the “State” must “suppress[]” evidence by failing to disclose it to the defendant. Strickler v. Greene, 527 U.S. 263, 281–82 (1999). Second, the undisclosed evidence must favor the defendant by helping to show the defendant’s innocence or to impeach an unfavorable witness. See id. at 280 (citing United States v. Bagley, 473 U.S. 667, 676 (1985)). Third, the nondisclosure must “prejudice” the defendant. Id. at 282. This prejudice exists if the undisclosed evidence is “material”—meaning that there is a “reasonable probability” that the trial outcome would have changed if the prosecution had turned over the evidence. Turner v. United States, 582 U.S. 313, 324 (2017) (citation omitted); Kyles v. Whitley, 514 U.S. 419, 434 (1995). According to Thurman, Hardin has not met any of these elements.

Starting with the first element, Thurman concedes that he did not disclose his observation notes to prosecutors or the defense. But he suggests that “there is zero evidence of bad faith” because he acted pursuant to his crime lab’s official policy and because the district court held that no jury could find that he knowingly gave false testimony. Appellant’s Br. 41–42; Clark, 2024 WL 55862, at *9. Even if Thurman were right on this factual claim, he gets the law wrong. Brady does not require officials to act with a nefarious intent when they fail to disclose apparently exculpatory evidence. See Strickler, 527 U.S. at 282. Rather, prosecutors can violate Brady not only though malicious concealment but also through honest mistakes. See United No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 9

States v. Agurs, 427 U.S. 97, 110 (1976). And under our circuit precedent, a Brady-derived obligation applies not just to prosecutors but also to police officers—so long as “the exculpatory value” of the undisclosed evidence should have been “apparent” to the officers. Moldowan v. City of Warren, 578 F.3d 351, 382–89 (6th Cir. 2009) (citation omitted). Contrary to Hardin’s claim, this precedent does not require a § 1983 plaintiff to show that an officer acted in “bad faith.” Id. at 388.

Turning to the second and third elements, Thurman argues that the district court wrongly held that the “exculpatory” value of the purported inconsistency between his observation notes and his testimony should have been “apparent” to him. Clark, 2024 WL 55862, at *11; Appellant’s Br. 38–41. And he argues that the court should not have found the notes “material” given that it rejected Hardin’s fabricated-evidence claim. Clark, 2024 WL 55862, at *12. Thurman reasons that the notes could not have created a reasonable probability of an acquittal (for Brady purposes) if they would not allow a jury to reject Thurman’s finding that Hardin’s hair sample and the hair found on Warford were similar (for fabricated-evidence purposes). Appellant’s Br. 44–45.

Both arguments suffer from the same problem: we lack jurisdiction over them under our precedent. Typically, we have jurisdiction only over appeals from “final decisions” that end a case in the district court. 28 U.S.C. § 1291; see Mohawk Indus., Inc. v. Carpenter, 558 U.S. 100, 106 (2009). But the Supreme Court’s “collateral order doctrine” allows a party to appeal an issue apart from a case’s merits if waiting to appeal until the end of the case would effectively bar the party from seeking appellate review of that issue. Mohawk, 558 U.S. at 106. And a denial of qualified immunity falls within this collateral-order doctrine because qualified immunity protects against the need to participate in the suit at all. See Mitchell v. Forsyth, 472 U.S. 511, 526–27 (1985). It does not just protect against a damages award. See id. A state actor may thus immediately appeal such a denial to the extent that it turns on an issue of law. Id. at 530.

This right to appeal, though, gives us the power to review only certain types of legal questions. We may review the primary question involved in any qualified-immunity appeal: Did the defendant’s actions as alleged violate “clearly established” law? Id. at 528. And we may No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 10

review abstract legal questions about the merits of the constitutional claims at issue. See Ashcroft v. Iqbal, 556 U.S. 662, 673–75 (2009); Hartman v. Moore, 547 U.S. 250, 257 n.5 (2006). But we cannot review the fact-specific (if admittedly legal) question about whether the evidence in the summary-judgment record sufficed to permit a reasonable jury to find some historical fact relevant to the claims. See Johnson v. Jones, 515 U.S. 304, 313 (1995); Gillispie v. Miami Township, 18 F.4th 909, 915–16 (6th Cir. 2021). We thus must distinguish unreviewable “evidence sufficiency” questions (about whether the record would allow a jury to find some fact) from reviewable “purely legal” questions (about what the caselaw clearly established or what a claim requires). Heeter v. Bowers, 99 F.4th 900, 909 (6th Cir. 2024). And we have often bemoaned the difficulty of deciding whether an appeal falls on the legal or factual side of this line. See id.

Thurman’s appeal concerns an even more nuanced question sitting between these extremes. Thurman does not raise a pure evidence-sufficiency challenge about whether the evidence would allow a reasonable jury to find some historical fact. But Thurman also does not raise an abstract legal challenge about the meaning of the Due Process Clause or the Supreme Court’s Brady decision. Rather, Thurman challenges the district court’s answer to what we have described as a “mixed” question of law and fact: Would the historical facts (when interpreted in Hardin’s favor) rise to the level required to meet the “favorability” and “materiality” elements of Hardin’s Brady claim? See Williams v. Mehra, 186 F.3d 685, 690 (6th Cir. 1999) (en banc); see also Gregory v. City of Louisville, 444 F.3d 725, 743–45 (6th Cir. 2006).

Does the collateral-order doctrine give us jurisdiction over these mixed questions of law and fact? It depends. We review some types of mixed questions in interlocutory appeals. For example, we routinely consider whether the historical facts in the summary-judgment record (read in the light most favorable to the plaintiff) rise to the level required to show that an officer used excessive force or had probable cause under the Fourth Amendment. See Heeter, 99 F.4th at 910–11, 912–15; Meadows v. City of Walker, 46 F.4th 416, 422–24 (6th Cir. 2022); Legenzoff v. Steckel, 564 F. App’x 136, 140 (6th Cir. 2014); see also Plumhoff v. Rickard, 572 U.S. 765, 772–73 (2014); Scott v. Harris, 550 U.S. 372, 381 n.8 (2007). And we have considered whether the historical facts in the summary-judgment record (again read in the light most favorable to the No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 11

plaintiff) show that a state official acted with deliberate indifference to a prisoner’s right to medical care under the Eighth Amendment. See Williams, 186 F.3d at 690.

Yet we have treated mixed questions in this Brady context as unreviewable at this stage. See Gregory, 444 F.3d at 743–44. In Gregory, a plaintiff who had been wrongly convicted of two rapes sued several investigating officers. 444 F.3d at 735–36. Among other claims, the plaintiff alleged that one officer had violated Brady by failing to disclose a similar rape that had occurred while the authorities were holding the plaintiff in custody. Id. at 736. The officer immediately appealed the district court’s denial of qualified immunity on this Brady claim. Id. at 736–37. He argued, among other things, that the existence of the similar rape was not “exculpatory” under Brady. Id. at 743. But we held that we lacked jurisdiction over this argument. Id. at 743–44. We reasoned that undisclosed evidence counts as “exculpatory” under Brady if it is “material” to the plaintiff’s “guilt or punishment.” Id. We then explained that “materiality under Brady is a mixed question of law and fact for the jury.” Id. at 744. And because the district court had held that a fact dispute existed over “the materiality of the [similar] rape,” we found that the officer raised only evidence-sufficiency challenges about these exculpatory and materiality issues. Id.

Gregory forecloses Thurman’s challenges here. He claims that his observation notes were not exculpatory and that his failure to disclose the notes did not prejudice Hardin because they were not material. But the district court held that a reasonable jury could find for Hardin on both of these elements of his Brady claim. Clark, 2024 WL 55862, at *11–12. Just as in Gregory, the elements “are beyond our reach at this juncture.” Moldowan, 578 F.3d at 389.

In response, Thurman does not try to distinguish Gregory’s holding that we treat mixed Brady questions like unreviewable evidence-sufficiency questions. He instead argues that we still have jurisdiction over his arguments under the Supreme Court’s Scott decision. See 550 U.S. at 380. There, the Court disregarded the plaintiff’s (and circuit court’s) version of what occurred during a high-speed chase because that version “blatantly contradicted” a video recording of the chase. Id. at 376, 380–81. But we have since described Scott as allowing us to consider an appeal challenging a § 1983 plaintiff’s version of evidence-sufficiency questions only in “exceptional circumstances,” such as those involving undisputed video evidence. No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 12

Gillispie, 18 F.4th at 916 (quoting Adams v. Blount County, 946 F.3d 940, 948 (6th Cir. 2020)). Here, nothing in the record undisputedly undermines the district court’s holdings that Thurman’s observation notes were exculpatory (because they clashed with his final report) and material (because they concerned critical evidence). We thus lack jurisdiction over Thurman’s challenges to those holdings.

B

Even if Thurman’s observation notes were favorable to Hardin and material to his defense, Thurman next makes the standard qualified-immunity argument over which we have jurisdiction: he asserts that the caselaw at the time of Hardin’s 1995 trial did not “clearly establish[]” that Brady’s duty of disclosure could apply to forensic scientists. Mitchell, 472 U.S. at 528. On this question, too, our precedent compels us to reject Thurman’s argument—this time on the merits.

Qualified immunity bars courts from imposing monetary liability on officers under § 1983 unless their conduct violated “clearly established” law. District of Columbia v. Wesby, 583 U.S. 48, 62–63 (2018) (citation omitted). To qualify as clearly established, a legal rule must have “a sufficiently clear foundation in then-existing precedent.” Id. at 63. The rule meets this test if “every reasonable” officer would agree that the rule applied to the situation that a defendant confronted and prohibited the defendant’s actions. Id. In other words, the precedent at the time of those actions must have rendered it “beyond debate” that the actions violated the law. Kisela v. Hughes, 584 U.S. 100, 104 (2018) (per curiam) (citation omitted).

Typically, these fair-notice principles will compel plaintiffs to identify the legal rule on which they rely with a “high ‘degree of specificity.’” Wesby, 583 U.S. at 63 (citation omitted). Plaintiffs have identified a clearly established legal rule at “too high a level of generality” if a court cannot “immediately” say that the defendant’s conduct violated the rule. Beck v. Hamblen County, 969 F.3d 592, 599 (6th Cir. 2020) (quoting Wesby, 583 U.S. at 64). When it is debatable how a general rule applies to a defendant’s specific conduct, a court cannot conclude that a reasonable officer should have known that the conduct violated the law. See Wesby, 583 U.S. at 63–64. No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 13

Thurman claims that Hardin cannot satisfy this qualified-immunity test. At the outset, he concedes that, by 1990, courts had clearly established the legal rule that forensic scientists cannot “deliberately” fabricate incriminating evidence or conceal exculpatory evidence in bad faith. Moldowan, 578 F.3d at 397; see Arizona v. Youngblood, 488 U.S. 51, 57−58 (1988); Spurlock v. Satterfield, 167 F.3d 995, 1005–07 (6th Cir. 1999). But the district court rejected Hardin’s fabricated-evidence claim. Clark, 2024 WL 55862, at *9. According to Thurman, this holding shows that he did not act in bad faith and thus that the clearly established rule does not apply.

That brings us to the question at the crux of the parties’ debate: do forensic scientists have a Brady-like duty of disclosure even absent bad faith? Thurman argues that Brady’s generic legal rule reads the right at “too high a level of generality.” Beck, 969 F.3d at 600. Even if prosecutors had an absolute duty of disclosure in the 1990s, Thurman reasons, that rule does not establish that everyone else connected to the investigation (including forensic serologists) also had such an absolute duty. Hardin counters that our caselaw has already rejected this claim.

Hardin is right. To be sure, Thurman correctly notes that Brady and the Supreme Court cases applying it have imposed the disclosure duty on “prosecutors,” not investigators like police officers or forensic scientists. Moldowan, 578 F.3d at 377; see Kyles, 514 U.S. at 437–38; Brady, 373 U.S. at 87. Yet we have held that pre-1990 cases clearly established that the police also have a “Brady-derived” duty to disclose material exculpatory evidence to the prosecution. Moldowan, 578 F.3d at 381–82; see Gillispie, 18 F.4th at 918 n.2; Jackson, 925 F.3d at 823−24. In Moldowan, for example, a state court had wrongly convicted the § 1983 plaintiff of a violent sexual assault, and he brought a Brady claim against (among others) an officer involved in the investigation. 578 F.3d at 363–67, 376. The plaintiff alleged that this officer violated Brady because he did not disclose an eyewitness’s “exculpatory statements” suggesting that other men had committed the assault. Id. at 376, 382. We held that “overwhelming” out-of-circuit precedent clearly established the extension of Brady’s no-fault regime to the police before 1990. Id. at 381–82.

Admittedly, this caselaw addressed claims against police officers rather than forensic scientists. But “every reasonable” scientist would have recognized that this extension of Brady covered them too. Wesby, 583 U.S. at 63. To be sure, one might have distinguished prosecutors No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 14

from others in the criminal investigation for Brady purposes. After all, a state official must answer essentially legal questions suited for lawyers when deciding whether a piece of evidence is material or exculpatory. See Moldowan, 578 F.3d at 379–81; id. at 402 (Kethledge, J., concurring in the judgment in part and dissenting in part). But once we make the leap beyond prosecutors, we see no reasonable argument why this leap should apply only to police officers in the field and not those in the lab. See Horn v. Stephenson, 11 F.4th 163, 171−73 (2d Cir. 2021); Brown v. Miller, 519 F.3d 231, 237−38 (5th Cir. 2008); Pierce v. Gilchrist, 359 F.3d 1279, 1299 (10th Cir. 2004).

An analogy confirms this point. In Spurlock, we found that the existing caselaw in 1990 clearly established that a police officer would violate due process by intentionally fabricating evidence. See 167 F.3d at 1005–06. In Gregory, we then relied on Spurlock to find that the caselaw clearly established this principle for forensic scientists. See 444 F.3d at 740, 744. Gregory saw “no reason to treat the intentional fabrication of a forensic report differently from the intentional fabrication of” a police report. Id. at 740; see Moldowan, 578 F.3d at 397. If we do not treat these two officials differently for purposes of a fabricated-evidence claim, why should we treat them differently for purposes of a Brady claim? In short, because we have held that it was clearly established in 1990 that Brady’s disclosure obligations covered police officers, we must now hold that it was clearly established that those obligations covered forensic examiners.

In response, Thurman claims that no case has clearly established that a forensic scientist might have a Brady duty to turn over “unfinished notes” that contain only initial “thoughts and impressions.” Appellant’s Br. 34. At this stage, however, our precedent requires us to assume what the district court held a reasonable jury could find about these notes: that their exculpatory nature was “apparent” and that they were material to Hardin’s innocence. Clark, 2024 WL 55862, at *11. And, as explained, it was clearly established in 1990 that a scientist could violate Brady by failing to disclose material exculpatory evidence. That rule suffices to rebut Thurman’s qualified-immunity defense under our caselaw. See Gillispie, 18 F.4th at 918 n.2; Jackson, 925 F.3d at 823−24; Moldowan, 578 F.3d at 381–82. Contrary to Thurman’s claim, No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 15

these cases do not require plaintiffs to identify a decision that addressed exactly the same type of Brady evidence.

* * *

For the foregoing reasons, we affirm in part and dismiss in part for lack of jurisdiction. No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 16

CONCURRENCE _________________

JANE B. STRANCH, Circuit Judge, concurring. The majority opinion correctly holds that we lack jurisdiction over Thurman’s argument that his notes were neither exculpatory nor material. I write separately to address the apparent inconsistency in this circuit’s precedent on the extent of our jurisdiction over mixed questions of fact and law in the context of appeals of denial of qualified immunity and to note that, even under a broader conception of this court’s jurisdiction, we would lack jurisdiction in this case.

In general, on appeals of qualified immunity, we have jurisdiction to the extent the issue on appeal turns on an issue of law. Mitchell v. Forsyth, 472 U.S. 511, 530 (1985). Where an officer’s dispute of facts is crucial to the appeal, we lack jurisdiction. Adams v. Blount Cnty., 946 F.3d 940, 951 (6th Cir. 2020). When presented with an appeal of a question that is traditionally a mixed question of fact and law, in most circumstances, we have jurisdiction provided either that we can resolve the question based on undisputed facts or that the appellant state official concedes the plaintiff’s version of the facts for the purposes of appeal. Williams v. Mehra, 186 F.3d 685, 689-90 (6th Cir. 1999) (en banc). Thus, we have considered appeals from denials of qualified immunity when the appeals turned on whether state officials were deliberately indifferent to a plaintiff’s medical needs, whether officials had probable cause, whether officials used excessive force, and any number of other mixed questions, provided that the appellant officials conceded the underlying facts. Id.; Heeter v. Bowers, 99 F. 4th 900, 908– 11 (6th Cir. 2024); Legenzoff v. Steckel, 564 F. App’x 136, 140 (6th Cir. 2014).

If we were to follow the same pattern in Brady cases, we would hold that we have jurisdiction to consider appeals of some denials of qualified immunity in which the appeal is founded on arguments that the allegedly suppressed evidence was not exculpatory or material. Under such a framework, if the appellant official conceded all of the plaintiff’s claims about what evidence the official was aware of, what evidence was suppressed, and what evidence was presented at trial, and conceded the factual inferences the district court drew therefrom, but argued that, nonetheless, such evidence was not material, we would have jurisdiction to hear the No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 17

official’s argument. But Gregory v. City of Louisville forecloses that possibility. Gregory held that we cannot consider any mixed questions of fact and law in the Brady context on an appeal from denial of qualified immunity, regardless of what factual concessions the appellant makes. 444 F.3d 725, 743–44 (6th Cir. 2006) (“[M]ateriality under Brady is a mixed question of law and fact for the jury. . . . The district court held that there exists a genuine issue of fact with respect to the materiality of the [suppressed evidence]. . . . Accordingly, this court lacks jurisdiction[.]”). I agree with Judge Murphy that this court’s holding in Gregory exits in some tension with the general pattern of our holdings in qualified immunity cases from a theoretical standpoint.

From a practical standpoint, however, it is important to note when analyzing such claims how narrow the gap between fact and law is when it comes to materiality. This, indeed, was a point we made in Gillespie v. Miami Twp., 18 F.4th 909, 915–19 (6th Cir. 2021). In that case, we did not automatically jump—as we might have been able to under Gregory—to the conclusion that, because the appellant’s claims were about materiality under Brady and the appeal was from a denial of qualified immunity, we lacked jurisdiction. Id. Rather, we rested our conclusion that we lacked jurisdiction on the alternative reasoning that the state official had not put forward arguments that appropriately rested on undisputed facts or facts conceded for the sake of appeal. Id. at 918. In doing so, we explained the extent to which a state official seeking to appeal a denial of qualified immunity must concede the facts:

An appellate court reviewing the denial of the defendant’s claim of immunity need not consider the correctness of the plaintiff’s version of the facts, nor even determine whether the plaintiff’s allegations actually state a claim. Because the scope of the defendant’s appeal is so circumscribed, we need look no further than the district court’s opinion, and we often may be able merely to adopt the district court’s recitation of facts and inferences. This court simply defers to the district court’s determinations of fact. And beyond those determinations, a defendant may not challenge the inferences that the district court draws from those facts, as that too is a prohibited fact-based appeal. *** If disputed factual issues are “crucial to” a defendant's interlocutory qualified immunity appeal, we may not simply ignore such disputes; we remain obliged to dismiss the appeal for lack of jurisdiction. Even if a defendant asserts arguments about whether the law was clearly established, if he fails to concede the most favorable view of the facts to the plaintiff and instead relies solely on his version No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 18

of the facts, this court cannot consider those otherwise valid arguments because he has failed to satisfy a crucial jurisdictional prerequisite.

Id. at 916–17 (cleaned up).

Thus, putting Gregory aside, even were we to consider mixed questions of fact and law on appeals of denials of qualified immunity, state officials challenging a district court’s holding that the suppressed evidence was material would need to walk a narrow tightrope to avoid losing jurisdiction. To maintain the appeal, the official would need to concede that what the plaintiff says he failed to disclose he failed to disclose. Id. More importantly, the official would need to concede all of the inferences the district court drew from that evidence—e.g., the district court’s finding that the evidence was inconsistent with evidence presented at trial or its inferences about what the evidence might have shown. Id. The official would need to argue that, even accepting all the district court’s factual inferences, the suppressed evidence was not material. This is not an easy claim to make. Thus, while Gregory closes a theoretical window for appeal, practically speaking, it forecloses a very narrow range of appeals.

Thurman’s claim in this case does not fall within that narrow range. Even adopting our general approach for jurisdiction over denials of qualified immunity, Thurman has not properly conceded the facts. To be sure, he repeatedly asserts that he does not dispute the facts. Reply Br. 8. But to do so, he must accept the district court’s conclusion that there were “inconsistencies between his observation notes and conclusion” that he failed to mention in his final report. R. 390, Order, PageID 34375. Yet almost immediately after denying reliance on factual disputes, Thurman argues that “there is no evidence that Thurman’s notes could contradict the conclusions in his final lab report.” Reply Br. 9. Thurman elaborates on his resistance to the district court’s finding of inconsistencies, arguing that the true source of his hair comparison conclusion was his microscopic, not naked eye, examination of the hairs. Reply Br. 9–10. He contends that his notes could not be inconsistent with his final report because they do not involve microscopic comparison. Reply Br. 9. Thurman’s final argument that expert testimony would be necessary to establish any inconsistency between his notes and his report completely rejects the district court’s finding of inconsistency. Reply Br. 9. No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 19

All of these arguments dispute the inferences drawn by the district court from the facts presented by the plaintiff. Under Gillespie, they amount to little more than disputes of the sufficiency of the evidence to establish that the notes were inconsistent with the final report. Gillespie, 18 F. 4th at 916-17; see also Barry v. O’Grady, 895 F.3d 440, 444–45 (6th Cir. 2018) (concluding that the arguments O’Grady presented as legal were actually factual disputes resting on O’Grady’s claim that there was “no evidence whatsoever” for some of Barry’s assertions of fact). Indeed, Thurman’s refusal to concede the facts is underscored by his immediate citation to Scott v. Harris for the proposition that “[w]hen opposing parties tell two different stories, one of which is blatantly contradicted by the record, so that no reasonable jury could believe it, a court should not adopt that version of the facts for purposes of ruling on a motion for summary judgment”— implicitly acknowledging that he is telling a different story from the one alleged by the plaintiff and adopted by the district court. Reply Br. 10 (citing 550 U.S. 372, 380 (2007)).

Because Thurman refuses to concede the relevant facts, even if we were to consider mixed questions of fact and law in this context as a theoretical matter, I would hold that we did not have jurisdiction over this particular claim. No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 20

MURPHY, Circuit Judge, concurring. In Brady v. Maryland, 373 U.S. 83 (1963), the Supreme Court held that the prosecution has a due-process duty to disclose evidence to the defense that meets the Court’s “favorability” and “materiality” elements. See id. at 87. Garr Keith Hardin asserts that Robert Thurman, a forensic serologist with the Kentucky State Police, violated Brady by failing to disclose his “observation notes” to the prosecution. Why? Those notes allegedly conflicted with Thurman’s later testimony that a hair found on Rhonda Sue Warford was “similar” to a sample of Hardin’s hair. Given our precedent, the majority opinion correctly resolves the two main Brady issues that this appeal presents. First, our caselaw holds that we lack jurisdiction over whether the summary-judgment record would allow a reasonable jury to find a Brady claim’s favorability and materiality elements. So we must refuse to consider Thurman’s arguments that his notes were neither exculpatory nor material. Second, our caselaw holds that plaintiffs may recover damages for Brady violations under 42 U.S.C. § 1983 without showing that a state defendant harbored any malicious intent when failing to disclose the identified evidence. So Thurman’s argument that he did not act in bad faith by failing to hand over his notes (and instead merely followed his lab’s policy) cannot rebut Hardin’s Brady claim. Although our cases require us to reach these two conclusions, I write this concurrence because I am not sure either is right.

Start with our jurisdictional holding. This holding depends both on the elements for Brady claims and on the rules for interlocutory appeals. As for Brady’s elements, it is now black-letter law that the Due Process Clause requires three things to establish a Brady violation. See Strickler v. Greene, 527 U.S. 263, 281–82 (1999). The government must have “suppressed” evidence. Turner v. United States, 582 U.S. 313, 324 (2017) (quoting Strickler, 527 U.S. at 282). This evidence must have favored the criminal defendant in some way. See id. at 323. The evidence might have been “exculpatory” because it would have helped show that the defendant did not commit the crime. Strickler, 527 U.S. at 282. Or it might have been “impeaching” No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 21

because it would have helped undercut the credibility of a witness who claimed the defendant did commit it. Id. Lastly, the government’s refusal to turn over the evidence must have harmed the defendant because the evidence was “material” to the outcome. Id. at 280, 282. Evidence is material if “there is a reasonable probability that, had the evidence been disclosed, the result of the proceeding would have been different.” Turner, 582 U.S. at 324 (quoting Cone v. Bell, 556 U.S. 449, 469–70 (2009)).

As for the interlocutory-appeal rules, a district court’s decision to deny summary judgment generally is not “final,” so a party usually may not appeal that decision. 28 U.S.C. § 1291. But the denial of qualified immunity in a § 1983 suit falls within an exception to this rule that goes by the “collateral-order doctrine.” See Mitchell v. Forsyth, 472 U.S. 511, 525–30 (1985). In Mitchell, the Supreme Court held that a state official may immediately appeal “an issue of law” embedded in a summary-judgment denial of qualified immunity. Id. at 530. The Court has since made clear that the collateral-order doctrine allows state officials to appeal only “abstract” legal questions, not “fact-based” ones. Ashcroft v. Iqbal, 556 U.S. 662, 674 (2009) (quoting Johnson v. Jones, 515 U.S. 304, 317 (1995)). An official, for example, may appeal the question whether our existing cases “clearly established” the alleged violation of a constitutional right. Plumhoff v. Rickard, 572 U.S. 765, 773 (2014). And an official may appeal the question whether a constitutional claim contains some disputed legal element. See Hartman v. Moore, 547 U.S. 250, 257 n.5 (2006). But the official may not appeal a pure evidence-sufficiency question about whether the record would allow the plaintiff to prove some fact. See Johnson, 515 U.S. at 313.

Our precedent has applied this appeal framework in a way that strictly limits the issues that a state actor may appeal in this Brady context. See Moldowan v. City of Warren, 578 F.3d 351, 389 (6th Cir. 2009); Gregory v. City of Louisville, 444 F.3d 725, 743–44 (6th Cir. 2006); see also Gillispie v. Miami Township, 18 F.4th 909, 917–18 (6th Cir. 2021). In Gregory, a police officer argued that certain undisclosed evidence was not “exculpatory.” 444 F.3d at 743. In response, we seemingly tied this favorability element of a Brady claim to its separate materiality element, noting that materiality raises a “mixed question of law and fact for the jury” to resolve. Id. at 744. And because the district court had held that a reasonable jury could find the evidence No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 22

material, we held that the officer raised a mere evidence-sufficiency challenge that fell outside our jurisdiction. See id. Since Gregory, we have treated the question whether undisclosed information could rise to the level of “material exculpatory evidence” as a factual determination that we lack jurisdiction to consider in an interlocutory appeal. Gillispie, 18 F.4th at 918; see Moldowan, 578 F.3d at 389.

I am dubious of this conclusion. To begin with, I know of no other area in which we lack jurisdiction over “mixed” questions about whether the record evidence (interpreted in the plaintiff’s favor) would meet the legal test for a constitutional right. Our en banc court instead described these “mixed questions as legal questions rather than as factual questions” for purposes of the collateral-order doctrine. Williams v. Mehra, 186 F.3d 685, 690 (6th Cir. 1999) (en banc). And while Williams addressed an Eighth Amendment claim, it did not suggest that we should adopt a right-by-right approach to decide whether we may review the district court’s application of a legal test to the historical facts in qualified-immunity appeals. Rather, Williams suggested that we categorically have jurisdiction over all such “mixed” constitutional questions. See id.

Our approach to Brady claims also conflicts with the Supreme Court’s approach to excessive-force claims. The Court has treated as a “legal issue[]” the question whether the evidence about a state actor’s conduct (interpreted in the plaintiff’s favor) violates the Fourth Amendment. Plumhoff, 572 U.S. at 772–73; Scott v. Harris, 550 U.S. 372, 381 n.8 (2007). I see no basis to distinguish that question (whether the force that a state actor used was excessive) from the question at issue here (whether the evidence that a state actor suppressed was material and exculpatory). If we have jurisdiction to review the first issue, why shouldn’t we have jurisdiction to review the second one?

Perhaps we could take a right-by-right approach on the ground that “[m]ixed questions are not all alike.” U.S. Bank Nat’l Ass’n v. Vill. at Lakeridge, LLC, 583 U.S. 387, 388 (2018). Some require courts to make “case-specific factual” determinations; others require courts to “expound on the law[.]” Id. at 396. The Supreme Court has thus held that the proper standard of review for a specific mixed question depends on whether the answer “entails primarily legal or factual work.” Id. But its caselaw in that standard-of-review context confirms that we likely have jurisdiction here. The Court presumes that all mixed questions “[i]n the constitutional No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 23

realm” trigger de novo review so that appellate courts can “mark[] out the limits” of a constitutional right “through the process of case-by-case adjudication[.]” Id. at 396 n.4 (quoting Bose Corp. v. Consumers Union of United States, Inc., 466 U.S. 485, 503 (1984)). Because the Court has treated mixed constitutional questions as legal for the standard of review, we likely should treat them as legal for our jurisdiction. Cf. Wilkinson v. Garland, 601 U.S. 209, 217–18 (2024).

The precedent on which Gregory relied confirms this point. Gregory cited United States v. Phillip, 948 F.2d 241, 250 (6th Cir. 1991), for the view that Brady materiality represents “a mixed question of law and fact for the jury.” 444 F.3d at 744. Phillip supports the first part of this quote since it treated Brady materiality as “a mixed question of law and fact[.]” 948 F.2d at 250. But Phillip rebuts the second part of the quote (“for the jury”). That decision arose on appeal from a criminal judgment, not in a civil case under § 1983. And it held that we review this mixed question (like other mixed Brady questions) under a “de novo” standard. Id.; see also Brooks v. Tennessee, 626 F.3d 878, 891 (6th Cir. 2010); United States v. Tarwater, 308 F.3d 494, 515 (6th Cir. 2002); Carter v. Bell, 218 F.3d 581, 591 (6th Cir. 2000). We do not normally apply de novo review to issues that we otherwise send to a jury for factual findings. So Phillip’s decision to give fresh review to these mixed questions suggests that they are legal ones ill-suited for a jury.

Admittedly, “[i]t is commonplace for the same mixed question of law and fact to be assigned to the court for one purpose, and to the jury for another.” United States v. Gaudin, 515 U.S. 506, 521 (1995). And maybe we could treat mixed Brady questions as factual for purposes of a § 1983 civil suit even though we treat them as legal for purposes of an appeal from a criminal judgment. Indeed, the Supreme Court in Gaudin held that a jury (not a judge) must resolve a similar “materiality” question in a criminal trial for making false statements. See id. at 522–23. As a result, if federal prosecutors had criminally charged Thurman with violating Brady by failing to disclose his observation notes, Thurman may well have possessed the right to have a jury decide Brady’s materiality element. See id. at 521 (citing 18 U.S.C. §§ 241–42).

Yet this fact does not change things. Gaudin’s holding rested on a criminal defendant’s Fifth and Sixth Amendment right to have a jury find “every element of the crime . . . beyond a No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 24

reasonable doubt.” Id. at 510. This constitutional concern disappears when we transition to civil cases under § 1983. In fact, Gaudin disclaimed addressing a court’s “power to resolve mixed- law-and-fact questions in civil cases[.]” Id. at 516–17. And this criminal-civil divide has mattered for other mixed questions of law and fact. Take probable cause. Gaudin opined that a jury would have to resolve whether the historical facts met the legal test for “probable cause” in a state official’s criminal trial for “depriving a person of constitutional rights under color of law[.]” Id. at 521; see United States v. Dukes, 779 F. App’x 332, 334 (6th Cir. 2019). After Gaudin, however, we have held that courts (not juries) should resolve this mixed probable-cause question in § 1983 suits. See Gerics v. Trevino, 974 F.3d 798, 805–06 (6th Cir. 2020). Gerics relied on the Supreme Court precedent applying de novo review to probable-cause questions in appeals from criminal judgments. Id. at 805 (discussing Ornelas v. United States, 517 U.S. 690, 695 (1996)). And, as noted, we have also applied de novo review to mixed Brady questions in appeals from criminal judgments. See Phillip, 948 F.2d at 250. So we likely should reach the same result as Gerics for these Brady questions in civil suits under § 1983.

To be sure, Gerics identified a lengthy history of courts rather than juries resolving the mixed probable-cause question in civil cases. See 974 F.3d at 805 n.6; see also id. at 808–10 (Thapar, J., concurring). I am not sure how civil courts have historically treated mixed questions like those raised under the Brady regime. Cf. Power v. Price, 16 Wend. 449, 455 (N.Y. 1836). Perhaps a different history could justify a different result. As of now, however, I see room to doubt our broad view that we lack jurisdiction to decide whether a certain set of facts (interpreted in the plaintiff’s favor) satisfies Brady’s “favorability” and “materiality” elements.

Even if we could review these “mixed” questions, Judge Stranch suggests that we would still lack jurisdiction because Thurman fails to take the historical facts in the light most favorable to Hardin. Yet those facts are largely undisputed: Thurman’s final report found the two hairs to be similar, whereas his notes (when taken in the light most favorable to Hardin) suggested that the hairs did not match in various ways. And Thurman accepts as true that his “notes may have been inconsistent with his report” in the identified ways. Reply Br. 10. After that, I am not sure what “inferences” are left other than the mixed-question inferences: Did the report’s inconsistencies rise to the level required to make the report both exculpatory and material? No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 25

And although I found Thurman’s briefing unclear, I ultimately read it to argue that the inconsistencies cannot make the notes exculpatory and material because the district court separately found that the report does not suffice to show that the two hairs were (in fact) different. Whether Thurman is right or wrong on these mixed questions, we likely should have jurisdiction to consider them. Cf. Barry v. O’Grady, 895 F.3d 440, 445–49 (6th Cir. 2018) (Sutton, J., dissenting).

Turn to our merits holding. Brady imposed a due-process duty of disclosure on prosecutors. See 373 U.S. at 87. In Moldowan, we interpreted the Due Process Clause as also imposing a “Brady-derived” duty on police officers, requiring them to deliver all “potentially exculpatory evidence” to prosecutors (but not defense counsel). 578 F.3d at 381. We added that criminal-defendants-turned-civil-plaintiffs can seek damages under § 1983 from officers who violate this Brady-like duty. See id. at 382–89. Yet we required these plaintiffs to establish different elements depending on the type of suppressed evidence. See id. If the evidence was “only ‘potentially useful’” to the defense, plaintiffs must show that officers acted in bad faith when failing to give it to prosecutors. See id. at 383–85 (relying on Arizona v. Youngblood, 488 U.S. 51, 57–58 (1988)). If, by contrast, the evidence’s “exculpatory value” was objectively “apparent,” plaintiffs can hold officers liable for this omission without the need to show that they acted with malicious intent. Id. at 388–89 (quoting California v. Trombetta, 467 U.S. 479, 489 (1984)). Moldowan refused to impose a bad-faith element for the latter type of evidence because an ordinary Brady claim does not require prosecutors to act with bad faith when failing to disclose material exculpatory evidence to the defense. See id. at 384; see also United States v. Agurs, 427 U.S. 97, 110 (1976). I agree that Moldowan’s holding clearly covers forensic examiners. Here, then, Hardin does not need to prove that Thurman acted in bad faith given the district court’s conclusion that a jury could find the “exculpatory value” of his observation notes “apparent.” Clark v. Louisville-Jefferson Cnty. Metro Gov’t, 2024 WL 55862, at *11 (W.D. Ky. Jan. 4, 2024).

That said, our holding solidifies a circuit split on this intent element. Moldowan itself recognized that “two of our sister circuits” have rejected Brady’s no-fault regime as applied to No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 26

§ 1983 claims against officers for their failure to disclose material exculpatory evidence to prosecutors. 578 F.3d at 383; see Porter v. White, 483 F.3d 1294, 1305–08 (11th Cir. 2007); Villasana v. Wilhoit, 368 F.3d 976, 980 (8th Cir. 2004). The Eighth Circuit requires a § 1983 plaintiff to prove bad faith—a phrase it defined to mean that an officer “intended to deprive [the plaintiff] of a fair trial” through “a conscious effort to suppress exculpatory evidence.” Villasana, 368 F.3d at 980 (citation omitted); see Helmig v. Fowler, 828 F.3d 755, 760, 762 (8th Cir. 2016). At least one other court has adopted a “bad faith” test since Moldowan. Owens v. Balt. City State’s Att’ys Off., 767 F.3d 379, 396 & n.6 (4th Cir. 2014). The Eleventh Circuit, by comparison, has held that a § 1983 plaintiff may not hold officers liable for negligently failing to disclose material exculpatory evidence to prosecutors, but it has left open whether the plaintiff may hold them liable for recklessly doing so. See Porter, 483 F.3d at 1308 & n.11. The Ninth Circuit has since adopted a recklessness test. See Tennison v. City & Cnty. of San Francisco, 570 F.3d 1078, 1087–89 (9th Cir. 2009). As far as I can tell, our requirement that the evidence’s exculpatory nature be “apparent” to officers stands alone because it adopts a negligence test asking whether officers “know or should know” of that exculpatory nature. Moldowan, 578 F.3d at 388 (emphasis added).

I also find it notable that Villasana has facts like this case. There, a jury convicted the § 1983 plaintiff of a rape he did not commit. 368 F.3d at 977–78. During the investigation, a forensic examiner sent only his lab report (not information underlying the report) to prosecutors pursuant to his lab’s official policy. 368 F.3d at 977. The plaintiff later used the undisclosed information underlying the report to overturn his conviction. Id. at 978. He then sued the examiner under § 1983 for failing to disclose this information earlier. Id. The Eighth Circuit first held that the prosecutor had not violated Brady. Id. at 978–79. More relevant here, it then decided that the plaintiff had to show the examiner’s bad faith to hold him liable on the Brady claim. Id. at 980. And the plaintiff could not do so because, among other reasons, the examiner had followed his “Crime Laboratory policy” when failing to disclose the information. Id. According to the court, “[a]cting in accordance with agency policy tends to show good faith rather than bad, whether or not the policy is sound.” Id. Here too, Thurman did not disclose his notes pursuant to the Kentucky State Police Crime Lab’s “policy to provide only final reports to No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 27

the prosecution[.]” Clark, 2024 WL 55862, at *11. Yet his good or bad faith does not matter in our court.

For the reasons that Judge Kethledge and Judge Wilkinson have explained, I am also not confident that we have joined the right side of this conflict. See Moldowan, 578 F.3d at 401–05 (Kethledge, J., concurring in the judgment in part and dissenting in part); Jean v. Collins, 221 F.3d 656, 659–63 (4th Cir. 2000) (en banc) (Wilkinson, J., concurring in the judgment). Section 1983 creates a civil remedy against a state actor who “subjects, or causes to be subjected,” another person “to the deprivation of any rights, privileges, or immunities secured by the Constitution[.]” 42 U.S.C. § 1983. The Supreme Court has held that any analysis of a § 1983 claim must proceed in two steps. At step one, courts must “‘identify the specific constitutional right’ at issue” in the case. Manuel v. City of Joliet, 580 U.S. 357, 370 (2017) (citation omitted). The “specific constitutional standard” will determine a constitutional claim’s initial validity. Graham v. Connor, 490 U.S. 386, 394 (1989). At step two, courts must identify the “elements of, and rules associated with,” that claim. Manuel, 580 U.S. at 370. This second step depends less on the meaning of the Constitution and more on the meaning of § 1983. And the Court has suggested that the claim should presumptively adopt the rules from the “most analogous” common-law tort at the time Congress passed the statute in 1871. Thompson v. Clark, 596 U.S. 36, 43 (2022); see Nieves v. Bartlett, 587 U.S. 391, 405 (2019); Heck v. Humphrey, 512 U.S. 477, 483–87 (1994).

Here, I see grounds at both steps to adopt an intent element for Hardin’s § 1983 claim. At step one, we must pinpoint the “specific constitutional right” that Hardin has invoked. Manuel, 580 U.S. at 370 (citation omitted). Because Hardin relies on Brady, I assume he means to assert a violation of the Fourteenth Amendment’s Due Process Clause. That clause makes clear that no “State” shall “deprive any person of life, liberty, or property, without due process of law[.]” U.S. Const. amend. XIV, § 1. As others have noted, see Porter, 483 F.3d at 1307–08; Jean, 221 F.3d at 660 (Wilkinson, J., concurring in the judgment), the Supreme Court has interpreted the verb “deprive” to require state officials to act with a level of intent beyond negligence before they can violate the Due Process Clause. See Daniels v. Williams, 474 U.S. 327, 330–31 (1986); see also Davidson v. Cannon, 474 U.S. 344, 347–48 (1986). In other No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 28

words, “negligently inflicted harm is categorically beneath the threshold of constitutional due process.” County of Sacramento v. Lewis, 523 U.S. 833, 849 (1998). So how could Thurman have violated due process if he only negligently failed to turn over the alleged evidence that would have purportedly led to Hardin’s acquittal?

To be fair, Brady and its progeny have read the Due Process Clause not to impose any intent element on prosecutors. See Kyles v. Whitley, 514 U.S. 419, 438 (1995); Brady, 373 U.S. at 87. And the Supreme Court has never reconciled Brady’s no-fault reading of the Due Process Clause with cases like Daniels or Lewis. Yet I see one potential basis for the reconciliation: no question exists that prosecutors intend to “deprive” criminal defendants of their “liberty” (or “property”) when they charge those defendants with crimes. U.S. Const. amend. XIV, § 1. One does not typically pursue a prosecution on accident. So perhaps this prosecutorial effort to take away a person’s liberty or property qualifies as a “deprivation under the Due Process Clause” even when prosecutors do not intend to suppress evidence during the prosecution. Daniels, 474 U.S. at 331. But police officers and forensic examiners do not make these intentional prosecutorial decisions. See Hartman, 547 U.S. at 261–62. So this logic may well not cover them.

Besides, even if step one of a § 1983 analysis does not require an intent element for this due-process claim, we still must proceed to step two. There, we must identify the statutory “elements of” the claim. Manuel, 580 U.S. at 370. And § 1983 undoubtedly requires more than the three-part test for Brady claims in a criminal case. See Strickler, 527 U.S. at 281–82. Consider causation. The statute requires defendants to have “subject[ed]” another person or “cause[d]” that person “to be subjected” to the violation of a constitutional right. 42 U.S.C. § 1983. This text likely requires plaintiffs to prove but-for and proximate causation by connecting their injuries to the defendant’s constitutional violation. See Marvaso v. Sanchez, 971 F.3d 599, 606 (6th Cir. 2020); Drumgold v. Callahan, 707 F.3d 28, 48–50 (1st Cir. 2013). So plaintiffs asserting Brady claims under § 1983 may well have to establish more than the materiality standard (a “reasonable probability” of a different outcome) that they must satisfy in a criminal case. Turner, 582 U.S. at 324 (citation omitted). As other courts have recognized, they may instead have to show “by a preponderance of the evidence” that the jury would have No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 29

acquitted them if the defendant had disclosed the exculpatory evidence. Drumgold, 707 F.3d at 49; see Gilliam v. Sealey, 932 F.3d 216, 238 (4th Cir. 2019); Rodriguez v. Woodall, 189 F. App’x 522, 527 (7th Cir. 2006). And they may well have to show that the specific prosecutor would have turned over this evidence to the defense if the prosecutor had known about it—since prosecutors (not police officers) ultimately decide whether the government must share evidence. Cf. Hartman, 547 U.S. at 262–64.

What about the other statutory elements? The Supreme Court has told us to start with the common-law tort that represents the “closest analogy” to these Brady claims. Nieves, 587 U.S. at 405 (quoting Heck, 512 U.S. at 484). We have yet to identify this tort. But our unpublished decisions have suggested that it is malicious prosecution. That caselaw has held that the so- called “Heck bar” applies to civil Brady claims, meaning that § 1983 plaintiffs must overturn their convictions before they may pursue their claims. See Hobbs v. Faulkner, 2020 WL 12933850, at *2 (6th Cir. June 9, 2020) (order); Ruiz v. Hofbauer, 325 F. App’x 427, 431 (6th Cir. 2009). And this Heck bar springs from the malicious prosecution’s “favorable-termination requirement” at common law. McDonough v. Smith, 588 U.S. 109, 117 (2019); see Thompson, 596 U.S. at 44.

If malicious prosecution provides the proper analogy, this Brady claim may well include an intent element as a statutory matter (if not a constitutional one). As the name of the tort implies, malicious prosecution required the defendant to have acted with a “malicious” intent. Thompson, 596 U.S. at 44 (quoting Thomas M. Cooley, Law of Torts 181 (1880)). The tortfeasor must have harbored “a purpose other than that of bringing an offender to justice.” Restatement (Second) of Torts § 668 (Am. L. Inst. 1977). The tortfeasor, for example, might not have believed in the plaintiff’s guilt or might have engineered the prosecution because of “hostility or ill will toward” the plaintiff. Id. § 668 cmt. d. In some respects, this common-law test resembles the intentional “bad faith” that some courts have required in this constitutional context. See Moldowan, 578 F.3d at 405 (Kethledge, J., concurring in the judgment in part and dissenting in part).

One last point: For qualified-immunity purposes, we have held that the existing precedent in 1990 clearly established a “Brady-derived” constitutional claim against law-enforcement No. 24-5061 Clark, et al. v. Louisville-Jefferson Cnty. Metro Gov’t, et al. Page 30

personnel. See id. at 381–82. But at least one other court has held that this Brady-like claim (with no intent element) was not clearly established around the same time. See Johnson v. City of Cheyenne, 99 F.4th 1206, 1232 (10th Cir. 2024). It instead suggested that the existing caselaw clearly established, at most, that officers can face liability for knowing or reckless failures to disclose material exculpatory evidence. See id.; see also Drumgold, 707 F.3d at 57 (Lynch, C.J., concurring in part, dissenting in part, and dissenting in the judgment). Our caselaw thus seemingly contradicts the decisions of other circuit courts on this separate qualified-immunity defense.

At day’s end, cases like Moldowan and Gregory bind us here. So I concur in the majority opinion. But I see room for debate over whether those decisions resolved these issues correctly.