I. Introduction

The following initial facts are undisputed. Kimberly Ken- nedy’s § 1983 action against Ridgefield City and Ridgefield Police Officer Noel Shields stems from events occurring on September 24, 1998, when a thirteen year-old neighbor, Michael Burns, shot and killed her husband, Jay Kennedy, and shot and severely wounded her. Earlier that same month, on September 6, Kennedy called the Ridgefield Police Department (“RPD”) and alleged that Burns had molested Kennedy’s nine-year-old daughter. RPD Officer Shields 2232 KENNEDY v. RIDGEFIELD responded to the call. Burns shot the Kennedys within approx- imately eight hours of first learning of the allegations against him. He has since been convicted of the premeditated murder of Jay Kennedy and the attempted premeditated murder of Kimberly.

At this early stage in the litigation, there are indeed facts which the parties dispute. However, because Shields contends that, even after resolving all issues of fact in Kennedy’s favor, she fails to demonstrate that he violated her constitutional rights, we present and consider the remaining facts, where appropriate, in a light most favorable to Kennedy.

During their initial meeting on September 6, Kennedy warned Shields of Michael Burns’s known, violent tenden- cies. She told Shields that the Burns family was unstable, that she had seen a lot of violence in their home, and described to Shields several violent incidents involving both Michael and his mother, Angela Burns. Kennedy told Shields that Michael had been involved in fights at school, had lit a cat on fire, had broken into his girlfriend’s house and attacked her with a baseball bat, and had thrown rocks at a building in downtown Ridgefield. After learning of Burns’s violent behavior, Shields assured Kennedy that she would be given notice prior to any police contact with the Burns family about her allega- tions.

Following that meeting, Shields forwarded his report to the Child Abuse and Intervention Center (“CAIC”). Shields had no further contact with Kennedy between September 6 and September 24, the night of the shooting. On several occasions, Kennedy inquired into the status of the investigation of Michael and reminded officers to notify her prior to any con- tact with the Burns family. In the interim, she and Shields both learned that Michael had been investigated for sending death threats to a classmate, though the investigation con- cluded he was not responsible. During her inquiries, Kennedy expressed concern for her safety and told the CAIC officer KENNEDY v. RIDGEFIELD 2233 handling the case that she was anxious to have the investiga- tion started.

On September 24, Kennedy called both Shields and the CAIC to inquire into the progress of the investigation. Ken- nedy left a message for Shields asking about the status of the alleged molestation case, and whether he had yet contacted Burns. After receiving Kennedy’s message when he arrived at work that afternoon, Shields called the CAIC to inquire into the status of the investigation. The officer responsible for the case was out, so Shields left his own message. Then, rather than calling Kennedy with an update, Officer Shields drove to the Burns residence. Shields claims he did so because the Burns house was on the way to the Kennedy’s, and if he could determine whether they had been contacted, he could continue to the Kennedy’s with more accurate information. At approxi- mately 5:00 p.m., Shields talked to Angela Burns, informing her and Michael of Kennedy’s allegations.

After speaking with Angela, Shields went to the Kennedy house. When he arrived, at approximately 5:15 p.m., Shields told Kennedy that he had informed Angela Burns of the molestation allegations. Kennedy became upset and asked Shields why he had contacted the Burns family prior to noti- fying her and told Shields that she feared for her safety. Offi- cer Shields assured her that the police would patrol the area around both her house and the Burns’s house that night to keep an eye on Michael.

After Shields left, Kennedy called a friend because she was very frightened of what Michael’s and his mother’s reactions would be. Shields had told her Angela was very angry after their conversation and that she and Michael had begun to yell at one another. Kennedy took no further action until about 10:00 p.m. that night when her husband returned from a hunt- er’s safety course. He had left their house to attend the course just as Shields had arrived that afternoon. The Kennedys decided to stay the rest of the night at home, in part because 2234 KENNEDY v. RIDGEFIELD of the late hour, and in part because Shields allegedly prom- ised to patrol the neighborhood. They planned to lock their doors and leave town early the next morning. But early on the morning of September 25, Michael Burns broke into the Ken- nedy house and shot both Jay and Kimberly Kennedy while they slept.

Kennedy filed suit against Shields and Ridgefield City, among others, in Clark County Superior Court asserting sev- eral state causes of action and a claim under 42 U.S.C. § 1983 and the Fourteenth Amendment. The case was removed to the United States District Court for the Western District of Wash- ington. On March 13, 2003, Shields and Ridgefield City moved for summary judgment. The court granted summary judgment to Defendants on Kennedy’s state law claims of negligent infliction of emotional distress and the tort of out- rage, and to Ridgefield City on her § 1983 “failure to train” claim.

However, the district court denied Shields’s motion for summary judgment based on qualified immunity. It concluded that, viewing the facts in a light most favorable to Kennedy, “a jury could find that Officer Shields unreasonably created a false sense of security in plaintiffs by agreeing to give plain- tiffs advanced notice of advising the Burns family of the alle- gation that Michael Burns sexually molested [Kennedy’s daughter], and assuring the plaintiffs of a neighborhood patrol.” Order, at 4-5. This interlocutory appeal followed.

II. Analysis

This case presents two legal issues. First, we must consider whether this Court has jurisdiction over Shields’s interlocu- tory appeal concerning his qualified immunity defense. If so, we must then determine whether Shields is entitled to such immunity.

We review de novo an interlocutory appeal from the denial of summary judgment based on qualified immunity. Wilkins KENNEDY v. RIDGEFIELD 2235 v. City of Oakland, 350 F.3d 949, 954 (9th Cir. 2003). In reviewing a summary judgment order in a § 1983 action where the district court determines that “the defendant’s alleged conduct violated the plaintiff’s clearly established constitutional rights[,] . . . we resolve all factual disputes in favor of the plaintiff . . . .” Cunningham v. City of Wenatchee, 345 F.3d 802, 807 (9th Cir. 2003).

A. Jurisdiction over Qualified Immunity Claims on Interlocutory Appeal

In response to Shields’s interlocutory appeal, Kennedy argues first that this court lacks jurisdiction. We disagree, and conclude we have jurisdiction to determine whether the trial court erred in holding Shields was not entitled to qualified immunity.

[1] As a general rule, interlocutory appeals from determina- tions of qualified immunity are permissible. In Mitchell v. Forsyth, 472 U.S. 511, 530 (1985), the Supreme Court held the denial of a defendant’s motion for summary judgment is immediately appealable where the defendant is a public offi- cial asserting the defense of qualified immunity, and the issue appealed concerns whether the facts demonstrated a violation of clearly established law.

[2] Kennedy correctly notes that the Court created an exception to this general rule in Johnson v. Jones, 515 U.S. 304 (1995). There, the Court held that “a defendant, entitled to invoke a qualified immunity defense, may not appeal a dis- trict court’s summary judgment order insofar as that order determines whether or not the pretrial record sets forth a ‘gen- uine’ issue of facts for trial.” Id. at 319-20. In ruling against Shields’s motion for summary judgment based on his claim of qualified immunity, the district court stated:

Viewed in a light most favorable to plaintiffs, a jury could find that Officer Shields unreasonably created 2236 KENNEDY v. RIDGEFIELD a false sense of security in plaintiffs by agreeing to give plaintiffs advance notice of advising the Burns family of the allegation that Michael Burns had sex- ually molested [Kennedy’s daughter], and assuring the plaintiffs of a neighborhood patrol. . . . In essence there is a question of fact as to whether or not there was justifiable reliance by plaintiffs on the alleged promises by Shields.

Order, at 4-5. Thus, the district court’s order observes that issues of fact remain.

[3] However, this does not suffice to deprive us of jurisdic- tion under Johnson. In a subsequent case, the Supreme Court explained:

Denial of summary judgment often includes a deter- mination that there are controverted issues of mate- rial fact, see Fed. Rule Civ. Proc. 56, and Johnson surely does not mean that every such denial of sum- mary judgment is nonappealable. Johnson held, sim- ply, that determinations of evidentiary sufficiency at summary judgment are not immediately appealable merely because they happen to arise in a qualified- immunity case. . . . Johnson reaffirmed that sum- mary judgment determinations are appealable when they resolve a dispute concerning an ‘abstract issu[e] of law’ relating to qualified immunity . . . typically, the issue whether the federal right allegedly infringed was ‘clearly established.’

Behrens v. Pelletier, 516 U.S. 299, 312-13 (1996); see also Knox v. Southwest Airlines, 124 F.3d 1103, 1107 (9th Cir. 1997) (“[W]e have jurisdiction over an interlocutory appeal from the denial of qualified immunity where the appeal focuses on whether the defendants violated a clearly estab- lished law given the undisputed facts, while we do not have jurisdiction over an interlocutory appeal that focuses on KENNEDY v. RIDGEFIELD 2237 whether there is a genuine dispute about the underlying facts.”).

Unlike the appeal in Johnson, we are neither asked nor required to look at the sufficiency of the evidence in support of the factual claims made by the parties, i.e., Shields’s con- tention that he did not create a false sense of security, and Kennedy’s insistence that he did. See Johnson, 515 U.S. at 313 (holding that some orders denying summary judgment, “though entered in a ‘qualified immunity’ case, determine[ ] only a question of ‘evidence sufficiency,’ i.e., which facts a party may, or may not, be able to prove at trial. This kind of order, we conclude, is not appealable.”).

[4] While the district court concluded that issues of fact remain, those disputed facts are not the basis of Shields’s interlocutory appeal before this court. Rather, Shields con- tends that, even after resolving the issues of fact in Kennedy’s favor, Kennedy will not have demonstrated that Shields vio- lated her clearly established, constitutional right. Because this question represents an “abstract issue of law relating to quali- fied immunity,” it falls within our jurisdiction on interlocu- tory appeal.

Assuming as true the facts adduced by Kennedy, then, we must determine whether Shields violated her constitutional rights and whether those rights were clearly established. Offi- cer Shields is entitled to qualified immunity unless we resolve both issues in the affirmative. We now turn to those questions.

B. Application of Qualified Immunity to Officer Shields

In Saucier v. Katz, 533 U.S. 194 (2001), the Supreme Court established a two-prong analysis for qualified immunity cases. First, a court must determine whether — resolving all disputes of fact and credibility in favor of the party asserting the injury — the facts adduced at summary judgment show that the offi- cer’s conduct violated a constitutional right. Saucier, 533 U.S. 2238 KENNEDY v. RIDGEFIELD at 201. If the court determines that the conduct did not violate a constitutional right, the inquiry is over and the officer is entitled to qualified immunity.

However, if the court determines that the conduct did vio- late a constitutional right, Saucier’s second prong requires the court to determine whether, at the time of the violation, the constitutional right was “clearly established.” Id. A right is clearly established if its “contours” are “sufficiently clear that a reasonable official would understand that what he is doing violates that right.” Id. (citing Anderson v. Creighton, 483 U.S. 635, 640 (1987)). Even if the violated right is clearly established, the Saucier Court recognized that, in certain situ- ations, it may be difficult for a police officer to determine how to apply the relevant legal doctrine to the particular cir- cumstances he or she faces. It held, therefore, that if an officer makes a mistake in applying the relevant legal doctrine, he or she is not precluded from claiming qualified immunity so long as the mistake is reasonable. That is, if “the officer’s mistake as to what the law requires is reasonable, . . . the offi- cer is entitled to the immunity defense.” Id. at 205.

1. First Prong: Did Shields Violate Kennedy’s Constitutional Rights?

Kennedy alleges that Shields violated her Fourteenth Amendment right to substantive due process by placing her in a known danger with deliberate indifference to her personal, physical safety.

[5] It is well established that the Constitution protects a citi- zen’s liberty interest in her own bodily security. See, e.g., Ingraham v. Wright, 430 U.S. 651, 673-74 (1977); Wood v. Ostrander, 879 F.2d 583, 589 (9th Cir. 1989). It is also well established that, although the state’s failure to protect an indi- vidual against private violence does not generally violate the guarantee of due process, it can where the state action “affir- matively place[s] the plaintiff in a position of danger,” that is, KENNEDY v. RIDGEFIELD 2239 where state action creates or exposes an individual to a danger which he or she would not have otherwise faced. DeShaney v. Winnebago County Dep’t of Soc. Serv., 489 U.S. 189, 197, 201 (1989); Wood, 879 F.2d at 589-90.1

This circuit first recognized such “danger creation” liability in Wood v. Ostrander, 879 F.2d 583 (9th Cir. 1989). In Wood, a state trooper determined that the driver of an automobile 1 The dissent suggests that this court created such liability in Wood by glossing DeShaney. See infra, at 2254-56. In fact, the “state-created dan- ger” doctrine predates DeShaney. See, e.g., White v. Rochford, 592 F.2d 381, 384 (7th Cir. 1979) (“[T]he complaint sufficiently alleged a depriva- tion of rights secured by the Constitution sufficient to state a claim under § 1983 . . . . [I]t is sufficient that the defendants left helpless minor chil- dren subject to inclement weather and great physical danger without any apparent justification.”); Bowers v. De Vito, 686 F.2d 616, 618 (7th Cir. 1982) (“If the state puts a man in a position of danger from private persons and then fails to protect him, it will not be heard to say that its role was merely passive; it is as much an active tortfeasor as if it had thrown him into a snake pit.”); Wells v. Walker, 852 F.2d 368, 370-71 (8th Cir. 1988) (“Circuit court decisions examining whether a particular individual, as dis- tinguished from the general public, is entitled to protection by the state from third-party harm generally recognize that the due process clause may be implicated in the following situation[ ] . . . when the state affirmatively places a particular individual in a position of danger the individual would not otherwise have been in.”) (citations omitted). See also David Pruess- ner, The Forgotten Foundation of State-Created Danger Claims, 20 Rev. Litig. 357 (2001) (tracing the modern doctrine to its roots in the statutory language and legislative history of the civil rights legislation originally enacted as the Ku Klux Klan Act of 1871, now codified as 42 U.S.C. § 1983). The oft-cited language of Deshaney, 489 U.S. at 201, is thus more reasonably understood as an acknowledgment and preservation of the doctrine, rather than its source. Moreover, the doctrine is not particular to our court. It is well estab- lished law in seven of our sister circuits. Butera v. District of Columbia, 235 F.3d 637, 651 (D.C. Cir. 2001); Dwares v. City of New York, 985 F.2d 94, 98-99 (2nd Cir. 1993); Kneipp v. Tedder, 95 F.3d 1199, 1201 (3d Cir. 1996); Kallstrom v. City of Columbus, 136 F.3d 1055, 1066-67 (6th Cir. 1998); Reed v. Gardner, 986 F.2d 1122, 1125 (7th Cir. 1993); Freeman v. Ferguson, 911 F.2d 52, 54-55 (8th Cir. 1990); Uhlrig v. Harder, 64 F.3d 567, 572 (10th Cir. 1995). 2240 KENNEDY v. RIDGEFIELD was intoxicated, arrested the driver and impounded the car. The officer’s actions allegedly left Wood, a female passenger, stranded late at night in a known high-crime area. Subse- quently, Wood accepted a ride from a passing car and was raped. This court held that Wood could claim section 1983 liability, since a jury presented with the above facts could find “that [the trooper] acted with deliberate indifference to Wood’s interest in personal security under the fourteenth amendment.” Id. at 588.

[6] Since Wood, this circuit has held state officials liable, in a variety of circumstances, for their roles in creating or exposing individuals to danger they otherwise would not have faced. See L.W. v. Grubbs, 974 F.2d 119 (9th Cir. 1992) (“Grubbs”) (holding state employees could be liable for the rape of a registered nurse assigned to work alone in the medi- cal clinic of a medium-security custodial institution with a known, violent sex-offender); Penilla v. City of Huntington Park, 115 F.3d 707 (9th Cir. 1997) (holding as viable a state- created danger claim against police officers who, after finding a man in grave need of medical care, cancelled a request for paramedics and locked him inside his house); Munger v. City of Glasgow, 227 F.3d 1082 (9th Cir. 2000) (holding police officers could be held liable for the hypothermia death of a visibly drunk patron after ejecting him from a bar on a bitterly cold night). These cases clearly establish that state actors may be held liable “where they affirmatively place an individual in danger,” Munger, 227 F.3d at 1086, by acting with “deliberate indifference to [a] known or obvious danger in subjecting the plaintiff to it,” L.W. v. Grubbs, 92 F.3d 894, 900 (9th Cir. 1996) (“Grubbs II”).2 2 We disagree with the dissent’s characterization of the factors for analy- sis our case law prescribes, infra at 2261-62. While it is proper to consider whether the conduct at issue was an affirmative act or an omission, whether it was directed toward the plaintiff specifically, and whether it was done with deliberate indifference to a known or obvious danger, we have never required, as the dissent suggests, infra at 2261, that the “gov- KENNEDY v. RIDGEFIELD 2241 a. Danger Affirmatively Created Due to State Action

[7] “In examining whether an officer affirmatively places an individual in danger, we do not look solely to the agency of the individual, nor do we rest our opinion on what options may or may not have been available to the individual. Instead, we examine whether the officer[ ] left the person in a situation that was more dangerous than the one in which they found him.” Munger, 227 F.3d at 1086. Thus, we ask first whether, as alleged, any affirmative actions by Shields placed Kennedy in danger that she otherwise would not have faced. Interpret- ing the facts in a manner most favorable to Kennedy, we con- clude they did.

[8] Shields drove to the Burns residence and notified the Burns family of the allegations against Michael. In doing so, he affirmatively created a danger to Kennedy she otherwise would not have faced, i.e., that Michael Burns would be noti- fied of the allegations before the Kennedys had the opportu- nity to protect themselves from his violent response to the

ernment’s act caused the harm” suffered by plaintiff. Instead, our “state- created danger” cases clearly contemplate § 1983 liability for the state actor who, though not inflicting plaintiff’s injury himself, has placed plaintiff in the harmful path of a third party not liable under § 1983. See United States v. Koon, 34 F.3d 1416, 1447-48 (9th Cir. 1994) (“The right which is established in these substantive due process cases is not the nar- row right to be protected from constitutional wrongs committed by third persons. Rather, because the individual has been placed in a dependent and helpless position, she is entitled to the broader right to be protected from harm. In . . . Grubbs, and in Wood, the third persons who inflicted the victims’ injuries, in fact, were not state actors. They were private citi- zens whose own actions could not have given rise to liability under . . . § 1983. The state actors — the defendants who failed to intervene, or who created the danger — were alone responsible for constitutional crimes or torts.”). Accordingly, we disagree with the dissent’s assertion, infra at 2263, that the state actor must be the “cause-in-fact of the plaintiff’s inju- ry.” Rather, the state actor need only have created the particularized risk that plaintiff might suffer such injury. 2242 KENNEDY v. RIDGEFIELD news. Like plaintiff’s supervisor in Grubbs, Shields created “an opportunity for [Burns] to assault [the Kennedys] that otherwise would not have existed,” Grubbs, 974 F.2d at 121.

The dissent’s assertion, infra at 2265, that “[n]otifying Michael Burns was an inevitable consequence of Kennedy’s allegations of child molestation” is an impermissible infer- ence from the facts.3 More importantly, it is beside the point. The only relevant question here is whether Shields, by informing Burns of Kennedy’s allegations without first warn- ing her as he had promised to do, realized the “inevitable con- sequence” about which the dissent speculates. We find that, in doing so, Shields affirmatively created an actual, particular- ized danger Kennedy would not otherwise have faced. The existence of this danger does not depend, as the dissent repeatedly suggests, infra at 2263-64 n.5, 2264, on a differ- ence of fifteen-minutes to which we give unwarranted consti- tutional magnitude. That Shields notified Kennedy of the danger he had created fifteen minutes before did not obviate or cure that danger; nor did it give Kennedy a reasonable opportunity to protect her family from it.

[9] In addition, we must accept Kennedy’s evidence that Shields assured her early in the evening of September 24 that, given the threat Michael posed, the police would patrol the 3 In fact record evidence clearly leads to the opposite inference. See, e.g., Appellee’s Supplemental Excerpts of the Record at 88 (recording deposi- tion testimony of a CAIC investigator: “Q: Do you receive any training as far as the timing when it’s best to contact an offender? A: At the end of the investigation. You need to have all your facts in order. Q: So by that you mean . . . that would be like the last step? A: Yes. Q: Why is that? A: Well, because you can’t tell when they’re lying to you. . . . Q: Is there a situation where you’ve been trained it’s good to contact the offender before the end of the investigation? A: The only time would be is if there was some sense of urgency, something that was emergent.”). In light of such evidence, the dissent’s speculation, infra at 2265 n.6, that the only reason for late notification is to allow a questioning officer to assess the offender’s credibility amounts to another impermissible inference drawn in Shields’s rather than Kennedy’s favor. KENNEDY v. RIDGEFIELD 2243 neighborhood that night. As in Grubbs, we do not rest our judgment that Shields affirmatively created a danger on that assurance alone, though in light of it, it is quite reasonable that the Kennedys decided late that night, when Mr. Kennedy returned from his class, to remain at home. Instead, as it did in Grubbs, Shields’s misrepresentation as to the risk the Ken- nedys faced was an additional and aggravating factor, making them more vulnerable to the danger he had already created. See Grubbs, 974 F.2d at 121 (“The Defendants also enhanced L.W.’s vulnerability to attack by misrepresenting to her the risks attending her work.”).4

b. Deliberate Indifference

We must decide the related issues of whether the danger to which Shields exposed the Kennedys was known or obvious, and whether he acted with deliberate indifference to it. See Bryan County v. Brown, 520 U.S. 397, 410 (1997) (“ ‘[D]eliberate indifference’ is a stringent standard of fault, requiring proof that a municipal actor disregarded a known or obvious consequence of his actions.”); Christie v. Iopa, 176 F.3d 1231, 1240 (9th Cir. 1999). Again, we look at the alleged facts in the light most favorable to Kennedy.

Kennedy has shown that, at their original meeting, she told Shields in detail of Michael Burns’s violent tendencies, including several incidents of what can only be described as alarming, aggravated violence, notably, lighting a cat on fire and assaulting his girlfriend with a baseball bat after breaking into her house. Additionally, she has testified that, after learn- ing of Burns’s violent behavior, Shields assured her that she 4 We note this court has already specifically rejected the “danger cre- ation” versus “danger enhancement” distinction the dissent raises, infra at 2266-67. See Penilla, 115 F.3d at 710 (“The critical distinction is not, as appellants allege, an indeterminate line between danger creation and enhancement, but rather the stark one between state action and inaction in placing an individual at risk.”). 2244 KENNEDY v. RIDGEFIELD would be given notice prior to any police contact with the Burns family. Kennedy also testified that between September 6 and 24, she left several messages with the police department and the CAIC in which she expressed continued fear for her family’s safety and refreshed her concern that she be given notice before the Burns family was notified in the course of the investigation.

On September 24, Shields knew that Michael was violent. Moreover, he knew that Michael had broken into his girl- friend’s house and beaten her with a baseball bat. On the facts alleged, it was obvious that Michael had a predilection for violence and was capable of the attack he in fact perpetrated on the Kennedys.5 Indeed, Burns’s attack was the very act Kennedy had repeatedly warned Shields of, and had sought to protect her family against. Thus, we are convinced that Shields knew that telling Burns about the allegations against him without forewarning the Kennedy’s would place them in a danger they otherwise would not have faced.

Kennedy also adduced sufficient evidence for us to con- clude that, if such evidence is accepted by the fact finder as true, Shields acted with deliberate indifference to the known and obvious danger we have just described. In Grubbs II, we clarified the mental state required in state-created danger cases. See 92 F.3d at 896. Despite its use of the term “deliber- 5 The dissent, infra at 2268, again appears to confuse the standard estab- lished in our case law by requiring foreseeability of the specific injury Burns in fact inflicted on the Kennedys, rather than foreseeability of the danger of such injury that Shields created. We have never required that, for a danger to exist, the exact injury inflicted by a third party must have been foreseeable. Instead, the state actor is liable for creating the foresee- able danger of injury given the particular circumstances. For example, in Wood, we did not speculate, nor require, that Trooper Ostrander foresee- ably knew Wood would in fact be raped by a passing motorist. We held he could be liable, however, for leaving Wood in a situation more danger- ous than the one she already faced, i.e., for stranding her alone in a known high-crime area at 2:30 a.m.. See Wood, 879 F.2d at 590. KENNEDY v. RIDGEFIELD 2245 ate indifference,” Wood had been interpreted to have estab- lished a “ ‘bare’ gross negligence” standard. Id. at 897-98. In Grubbs II, after surveying the standards of our sister circuits, we made clear that the standard in this circuit was not gross negligence but “deliberate indifference to a known, or so obvious as to imply knowledge of, danger.” Id. at 900. We explicitly said that such a mental state “is enough” — no more, no less. Moreover, we refused to parse it further, explaining, “[w]e have not added a requirement that the con- science of the federal judiciary be shocked by deliberate indif- ference, because the use of such subjective epithets as ‘gross’ ‘reckless’ and ‘shocking’ sheds more heat than light on the thought process courts must undertake in cases of this kind.”6 Id.

[10] Viewing the facts in the light most favorable to Ken- nedy, we find that, if accepted as true, they are sufficient to establish that Shields acted deliberately and indifferently to the danger he was creating. Kennedy warned Shields repeat- edly about Burns and requested that Shields notify her first so she could protect her family. With knowledge of Burns’s pro- pensity for violence and of Kennedy’s fear, and despite his promise to Kennedy to the contrary, Shields nevertheless noti- fied Burns first. Of all the possible actions he could take, and pursuant to no investigatory duties, he took the one most feared by Kennedy. His only explanation for his action is that it was a more convenient way in which to answer an adminis- trative phone message. Then, after notifying Burns, Shields allegedly reassured the visibly frightened Kennedy of increased security which was either never provided or plainly ineffective. Given the danger created by Shields that the Ken- 6 Citing language from our Grubbs II survey of other circuits, the dissent appears to suggest Shields was required to have a mental state closer to the specific intent of exposing Kennedy to the actual injury Burns inflicted. See infra at 2268-69. We disagree. Grubbs II requires no more and no less than “deliberate indifference” to the danger in question. 92 F.3d at 900. 2246 KENNEDY v. RIDGEFIELD nedys faced, we find such alleged, capricious behavior suffi- cient evidence of deliberate indifference.

2. Second Prong: Was the Right Violated Clearly Established?

We turn now to the second prong of Saucier, which Plain- tiff has the burden of establishing. See Sorrels v. McKee, 290 F.3d 965, 969 (9th Cir. 2002). We consider whether Kennedy has shown that the constitutional right violated by Shields was “clearly established” in September 1998. For the reasons below, we conclude she has.

To determine whether a right is clearly established, the reviewing court must consider whether a reasonable officer would recognize that his or her conduct violates that right under the circumstances faced, and in light of the law that existed at that time. Saucier, 533 U.S. at 202. As the Supreme Court has explained:

For a constitutional right to be clearly established, its contours must be sufficiently clear that a reasonable official would understand that what he is doing vio- lates that right. This is not to say that an official action is protected by qualified immunity unless the very action in question has previously been held unlawful . . . but it is to say that in the light of pre- existing law the unlawfulness must be apparent.

Hope v. Pelzer, 536 U.S. 730, 739 (2002) (citing Anderson v. Creighton, 483 U.S. 635, 640 (1987) (internal citations omit- ted). However, “[i]n order to find that the law was clearly established . . . we need not find a prior case with identical, or even ‘materially similar’ facts. Our task is to determine whether the preexisting law provided the defendants with ‘fair warning’ that their conduct was unlawful.” Flores v. Morgan Hill Unified Sch. Dist., 324 F.3d 1130, 1136-37 (9th Cir. 2003) (citing Hope, 536 U.S. at 740). KENNEDY v. RIDGEFIELD 2247 Thus, the specific, alleged conduct in this case need not have been previously and explicitly deemed unconstitutional, but existing case law must have made it clear that the conduct violated constitutional norms. This has been our consistent standard since Wood. See Wood, 879 F.2d at 592 (“[Defendant] seemingly suggests that this case can be dis- posed of if it does not bear a strict factual similarity to previ- ous cases finding liability. However this crabbed view of the good faith immunity principle cannot withstand analysis.”) (citing Anderson, 483 U.S. at 640).

[11] It is beyond dispute that in September 1998, it was clearly established that state officials could be held liable where they affirmatively and with deliberate indifference placed an individual in danger she would not otherwise have faced. This court first recognized the theory of state-created danger liability almost ten years before the events in this case in Wood. In the interim, we published three decisions explic- itly recognizing such liability under three distinct factual scenar- ios.7 See Grubbs, 974 F.2d 119; Koon, 34 F.3d 1416; Penilla, 7 The dissent, claiming to follow the requirements of Saucier, infra at 2271, attempts to show through an elaborate fact-matching exercise that none of our state-created danger cases clearly enough established the req- uisite notice. We consider the exercise misguided and, as discussed below, analytically flawed. An exact factual predicate case has never been required to find a right clearly established. Indeed, Flores made it clear that not even materially similar facts were necessarily required. See 324 F.3d at 1136-37. All a plaintiff need show is that a reasonable person would have understood from the case law that his actions would violate another’s constitutional rights. Hope, 536 U.S. at 739 (citing Anderson, 483 U.S. at 640). For excessive force cases, like Saucier and Brosseau v. Haugen, 543 U.S. 194 (2004), a somewhat more detailed analysis of the factual context is necessary because of what the Supreme Court deemed the “hazy border between excessive and acceptable force.” See Saucier, 533 U.S. at 205-6. For the case at hand, however, the factual scenarios in Woods, Grubbs, Koon, and Penilla were sufficient to clearly establish the following: a reasonable officer with Shields’s knowledge would have understood that informing Burns about Kennedy’s allegations — before adequately warning the Kennedys — would put them in greater danger than they otherwise would have faced that night. 2248 KENNEDY v. RIDGEFIELD 115 F.3d 707. Indeed, almost three years before the actions at issue in this case, we concluded “the law was clearly estab- lished that officers may be liable where they affirmatively place an individual in danger.” See Munger, 227 F.3d at 1086.8 We have explained before that the responsibility for keeping abreast of constitutional developments rests “squarely on the shoulders of law enforcement officials. Given the power of such officials over our liberty, and sometimes over our lives, this placement of responsibility is entirely proper.” Wood, 879 F.2d at 595 (quoting Ward v. County of San Diego, 791 F.2d 1329, 1332 (9th Cir. 1986)). We conclude that no reasonable officer in Shields’s position, knowing what he knew, could have concluded that Kennedy had no right not to be placed in physical danger by his deliberately indifferent action.

Indeed, even were we to engage in an examination of our case law with the finer resolution encouraged by the dissent, we conclude that, as to the state-creation of danger, this case is not “meaningfully distinguishable” from Grubbs. See Wood, 879 F.2d at 593. In Grubbs, a registered nurse working at a medium security custodial institution brought a § 1983 claim against her supervisors after she was allegedly raped and terrorized by a young male inmate. According to the plaintiff, her employer had told her she would not be working alone with violent sex offenders. Notwithstanding that repre- sentation, her employer subsequently allowed an inmate prone to violence against women to work with her unsupervised. The plaintiff, relying upon that representation, did not take all the precautions she might otherwise have taken, and was sub- sequently assaulted. 8 We note that the Fifth Circuit looking only to our decisions in Wood and Grubbs considered the state-created danger theory “clearly estab- lished” in this circuit under Saucier as early as 1993. See McClendon v. City of Columbia, 305 F.3d 314, 330, 324-25 (5th Cir. 2002) (identifying courts that had accepted “some version of this ‘state-created danger’ theo- ry”); id. at 328 n.10 (recognizing that “if this court had expressly adopted or rejected the state-created danger theory prior to [the incident date] that would, of course, be the end of our inquiry.”). KENNEDY v. RIDGEFIELD 2249 In Grubbs, as in this case, a state official affirmatively acted: supervisor Grubbs assigned a violent sex offender to work closely with L.W., and Officer Shields notified Burns, leaving Kennedy unable to protect her family. In Grubbs, as in this case, those state actions left plaintiffs exposed to the danger of the subsequent physical assault and injury they in fact suffered. And in both cases the plaintiff relied upon the state actor’s representation and did not take protective mea- sures she otherwise would have taken, and the state’s action made plaintiffs vulnerable to a particularized danger they would not have faced but for that action.

[12] Indeed, in this case, as in Grubbs, Shields used his “authority as a state . . . officer to create an opportunity for [Burns] to assault [Kennedy] that would not have otherwise existed.” Grubbs, 974 F.2d at 121. Moreover, Kennedy, like L.W., “is not seeking to hold Defendant[ ] liable for [Burns’s] violent proclivities. Rather, [she] seeks to make Defendant[ ] answer for [his] acts that independently created the opportu- nity for and facilitated [Burns’s] assault on her.” Id. at 122. At bottom Kennedy’s claim is exactly like L.W.’s, i.e., that a state actor “enhanced [her] vulnerability to attack by misrep- resenting to her the risks” she faced. Id. at 121. No reasonable officer in Shields’s position, knowing what he allegedly knew and what he must be charged with knowing, could have con- cluded otherwise than that Kennedy had a right not to be placed in obvious physical danger as a result of his deliber- ately indifferent action.

III. CONCLUSION

[13] Under Behrens, 516 U.S. at 312-13, we have jurisdic- tion to hear Shields’s interlocutory appeal regarding qualified immunity. On the merits, we conclude that, on this summary judgment record, Shields unreasonably violated Kennedy’s clearly established constitutional right. Under the state- created danger doctrine, a police officer may be liable for actions that create or increase a known or obvious danger to 2250 KENNEDY v. RIDGEFIELD an individual that he or she would otherwise not face. Because we hold that this doctrine was clearly established at the time the events of this case took place, and that Shields’s actions both created and aggravated the risk Plaintiff faced from Burns on the night of September 24, 1998, the district court’s denial of Shields’s motion for summary judgment based on qualified immunity is

AFFIRMED.

BYBEE, Circuit Judge, dissenting:

I vigorously part company with the majority’s conclusions that Shields created the danger that Kennedy faced and that he acted with deliberate indifference in doing so, thereby violat- ing her rights under the Due Process Clause of the Fourteenth Amendment. The majority’s conclusion is unsupported by the record and our own case law. The majority concludes that in the fifteen minutes between the time Officer Shields contacted Angela Burns and the time he advised Kim Kennedy of the contact, he deprived Kennedy of her due process rights. In so holding, the majority not only mangles the state-created dan- ger doctrine, it holds that its new rule was so clearly estab- lished that Officer Shields should have known he was violating the Constitution and, thus, has forfeited his qualified immunity.

We have never before recognized a state-created danger cause of action on facts remotely analogous to these. In the sixteen years since we introduced the state-created danger exception to DeShaney into our case law, we have approved its application on fewer than five occasions. In these cases, we have narrowly construed the exception to encompass only those claims in which the government’s action was directed at a specific plaintiff, rather than the public at large; the gov- ernment acted affirmatively, rather than simply failed to act; KENNEDY v. RIDGEFIELD 2251 the government’s act caused the harm, rather than merely increased the risk; and the government’s action constituted deliberate indifference to the known or obvious danger, rather than mere—or even gross—negligence. Ignoring these ele- ments, the majority today extends the state-created danger doctrine to a situation in which it cannot be said with any measure of confidence either that the government’s act caused the plaintiff’s harm or that the government acted with the req- uisite level of culpability.

Even if I thought Officer Shields had violated our state- created danger gloss on the Due Process Clause, the violation was surely not so obvious that he should have known at the time that he was violating Kennedy’s constitutional rights. Consequently, even assuming a constitutional violation, I would hold that Officer Shields is nonetheless entitled to qualified immunity. I respectfully dissent.1

I. BACKGROUND

The facts of this case are undeniably tragic. On September 6, 1998, Kennedy filed a complaint with the City of Ridge- field Police Department (“RPD”) accusing her thirteen-year- old neighbor, Michael Burns, of sexually molesting her nine- year-old daughter. Officer Shields was dispatched to Kenne- dy’s home to record the complaint.

Kennedy recalls talking with Officer Shields about the instability of the Burns family. She alleges that she informed Shields that the Burns family “had bad tempers” and that Michael was in trouble all the time, including one unfruitful investigation for allegedly sending a death threat to a class- mate; he also once threw rocks at his stepfather’s building. On another occasion, Michael reportedly lit a cat on fire, and later unlawfully entered his girlfriend’s house “and went after her 1 Although I dissent on the merits, I agree with the majority’s conclusion that we have jurisdiction to hear this interlocutory appeal. 2252 KENNEDY v. RIDGEFIELD with a baseball bat” after she broke up with him. On the basis of this alleged misconduct, Kennedy requested prior notifica- tion before the Burns family was informed of her allegations.

Following her initial complaint, Kennedy repeatedly con- tacted the RPD—at least six times during the eighteen days following her complaint—regarding the status of the investi- gation. On September 24, Kennedy called Officer Shields directly to determine whether the Burns family was aware of her allegations. Unable to reach Shields by phone, she left a message. In response to her inquiry, Shields proceeded to the Burnses’ home to ascertain whether the family had been noti- fied. Shields was greeted by Angela Burns (Michael Burns’s mother) and Shields asked her whether she had received a phone call or visit from the Child Abuse and Intervention Center (“CAIC”). Angela Burns inquired as to the reason for his question, and Shields advised her of the allegations.

Immediately following this meeting, Shields drove directly to Kennedy’s residence—located approximately one block away—and informed her that Angela Burns had been notified of her allegations. Kennedy alleges that she expressed fear regarding Michael Burns’s possible reaction. She further alleges that, in response to her expressions, Officer Shields promised to patrol the area that night to watch for Michael. After discussing the matter with her husband, Kennedy chose to remain in her home that evening and leave town the follow- ing morning. Michael Burns entered the Kennedy home that night, shot and killed Jay Kennedy, and seriously wounded Kim Kennedy. She now brings this action against Officer Shields, claiming that his conduct violated her rights under the Due Process Clause of the Fourteenth Amendment.

II. SAUCIER TWO-STEP

As the majority notes, the Supreme Court’s opinion in Sau- cier v. Katz, 533 U.S. 194 (2001), provides the framework for our analysis of this § 1983 suit. Under this framework, if a KENNEDY v. RIDGEFIELD 2253 defendant claims qualified immunity, we must make two dis- tinct inquiries: a “constitutional inquiry” and a “qualified immunity inquiry.” See Estate of Ford v. Ramirez-Palmer, 301 F.3d 1043, 1049 (9th Cir. 2002).

Officer Shields claims that he is entitled to qualified immu- nity from Kennedy’s suit. Accordingly, Saucier instructs that we must first determine whether, “[t]aken in the light most favorable to the party asserting the injury . . . the facts alleged show the officer’s conduct violated a constitutional right.” Saucier, 533 U.S. at 201. “[I]f a violation could be made out on a favorable view of the parties’ submissions, the next, sequential step is to ask whether the right was clearly estab- lished . . . in light of the specific context of the case” such that “it would be clear to a reasonable officer that his conduct was unlawful in the situation he confronted.” Id. at 201-02 (citing Wilson v. Layne, 526 U.S. 603, 615 (1999)).

The majority concludes that Kennedy’s allegations permit a jury to find that Officer Shields’s conduct deprived her of due process as guaranteed by the Fourteenth Amendment on the theory that Shields affirmatively created the danger that injured her and took her husband’s life. The majority holds, in addition, that Officer Shields is not entitled to qualified immunity for this violation. I disagree on both accounts. To explain my disagreement on the first point, it is worth briefly outlining this Court’s state-created danger doctrine.2

A. State-Created Danger Doctrine

As the majority observes, the state-created danger doctrine is said to trace its jurisprudential pedigree in this Circuit to the 2 The majority devotes a lengthy footnote to establishing the pre- DeShaney existence and the current prevalence of the state-created danger doctrine. I do not dispute that this doctrine is well established, merely its application to this case. On this note, the cases cited by the majority in its footnote support my view of this doctrine; see footnote 9, infra. 2254 KENNEDY v. RIDGEFIELD Supreme Court’s opinion in DeShaney, perhaps best known for Justice Blackmun’s exclamation, “Poor Joshua!” DeShaney v. Winnebago County Dep’t of Soc. Servs., 489 U.S. 189, 213 (1989) (Blackmun, J., dissenting). Declining to find a due process violation where local officials failed to ade- quately respond to complaints that four-year-old Joshua was being abused by his father, the Court held that the Constitu- tion does not require the state to protect the life, liberty, and property of its citizens against invasion by private actors. Rather, the Due Process Clause “is phrased as a limitation on the State’s power to act, not as a guarantee of certain minimal levels of safety and security.” Id. at 195. The Court observed,

Like its counterpart in the Fifth Amendment, the Due Process Clause of the Fourteenth Amendment was intended to prevent government from abusing its power, or employing it as an instrument of oppres- sion[.] Its purpose was to protect the people from the State, not to ensure that the State protected them from each other. The Framers were content to leave the extent of governmental obligation in the latter area to the democratic political processes.

Consistent with these principles, our cases have recognized that the Due Process Clauses generally confer no affirmative right to governmental aid, even where such aid may be necessary to secure life, lib- erty, or property interests of which the government itself may not deprive the individual. . . . [I]t follows that the State cannot be held liable under the Clause for injuries that could have been averted had it cho- sen to provide them.

Id. at 196-97 (internal quotation marks and citations omitted).

We have noted two distinct exceptions to the general rule that the state has no affirmative duty to protect persons from violence inflicted by private actors: (1) the “special relation- KENNEDY v. RIDGEFIELD 2255 ship” exception, stemming from a custodial relationship between the state and the victim; and (2) the “danger cre- ation” exception, stemming from “affirmative conduct on the part of the state in placing the plaintiff in danger.” L.W. v. Grubbs, 974 F.2d 119, 121 (9th Cir. 1992) (“Grubbs I”). The former emanates from language in DeShaney itself. DeShaney, 489 U.S. at 199-200 (“[W]hen the State takes a person into its custody and holds him there against his will, the Constitution imposes upon it a corresponding duty to assume some responsibility for his safety and general well- being.”). The latter, more amorphous, doctrine of “state- created danger” was developed by lower courts in response to the DeShaney Court’s observation that Winnebago County neither helped to create the dangers that Joshua faced nor ren- dered him more vulnerable to those dangers. DeShaney, 489 U.S. at 201 (“While the State may have been aware of the dangers that Joshua faced . . . it played no part in their cre- ation, nor did it do anything to render him any more vulnera- ble to them.”).

1. Ninth Circuit Cases

We established the state-created danger theory four months after DeShaney was published by recognizing a cognizable due process violation where the plaintiff alleged that she was raped after a state trooper impounded the vehicle in which she was riding, ejected her from the vehicle, and left her stranded in a high-crime area in the middle of the night. Wood v. Ost- rander, 879 F.2d 583 (9th Cir. 1989). In holding that Wood raised a triable issue of fact as to whether Trooper Ostrander’s conduct violated her substantive due process rights, we drew a distinction between facts demonstrating that police action created the danger to the person and facts demonstrating a danger that existed without police action. Wood, 879 F.2d at 589-90. Relying on Deshaney, we held that a substantive due process claim could be stated when police create the danger to an individual. We reasoned that “[t]he fact that Ostrander arrested [the driver], impounded his car, and apparently 2256 KENNEDY v. RIDGEFIELD stranded Wood in a high-crime area at 2:30 a.m. distinguished Wood from the general public and triggered a duty of the police to afford her some measure of peace and safety.” Id. at 590. Reversing the district court’s summary judgment for defendants, we concluded that the plaintiff’s allegations dem- onstrated “an assertion of government power which . . . tends to show a disregard for [her] safety amounting to deliberate indifference.” Id. at 588.

We further defined the contours of the state-created danger theory in Grubbs I, in which a registered nurse employed by the state of Oregon at a medium-security custodial institution brought suit against state prison officials after she was bat- tered, kidnapped, robbed, and raped by an inmate with known violent propensities. 974 F.2d at 120. The plaintiff alleged that she was led to believe that she would not have to work alone with residents who were known violent sex offenders. Id. Finding a cognizable due process violation, we empha- sized that the state had knowledge of the inmate’s dangerous propensities, and it affirmatively assigned him a job in which he would work alone with the plaintiff. Id. at 121. We con- cluded that the defendants, like the officer in Wood, “used their authority as state correctional officers to create an oppor- tunity for [the inmate] to assault [the plaintiff] that would not otherwise have existed.” Id. (emphasis added). We further observed that the defendants “enhanced [the plaintiff’s] vul- nerability to attack by misrepresenting to her the risks attend- ing her work”; namely, by leading her to believe that she would not be assigned to work alone with any inmates who were known violent sex offenders. Id.

Contrary to the majority’s suggestion, the “enhanced vul- nerability” that ensued from the state’s misrepresentation of the risks that the nurse would face in her employment did not, by itself, give rise to the due process violation recognized in Grubbs I. Maj. Op. at 2243, 2248-49. Indeed, under DeShaney, it is, at the very least, questionable whether a state’s failure to fully apprise an individual of the risks attend- KENNEDY v. RIDGEFIELD 2257 ing her employment can ever constitute an affirmative exer- cise of state power sufficient to give rise to a due process violation. See DeShaney, 489 U.S. at 201-02 (suggesting that the affirmative exercise of state power, as opposed to mere inaction, is the minimum threshold requirement necessary to establish a due process violation, and declining to find such affirmative exercise even in the context of an elaborate and exclusive system of child-protection services). Rather, Grubbs I more accurately stands for the proposition that, in order to state a claim based on state-created danger, the state must affirmatively play a part in creating the danger. See Grubbs I, 974 F.2d at 121 (“The ‘danger creation’ basis for a claim . . . necessarily involves affirmative conduct on the part of the state in placing the plaintiff in danger.”); see also Munger v. City of Glasgow Police Dep’t, 227 F.3d 1082, 1086 (9th Cir. 2000) (noting that the court in a state-created danger case “must determine whether [the state] did in fact affirmatively place [the plaintiff] in danger”).

In a second appeal in Grubbs, we addressed the level of culpability required to prevail under a state-created danger theory. See L.W. v. Grubbs, 92 F.3d 894 (9th Cir. 1996) (“Grubbs II”). Explicitly rejecting a “gross negligence” stan- dard, we held that “the plaintiff must show that the state offi- cial participated in creating a dangerous situation, and acted with deliberate indifference to the known or obvious danger in subjecting the plaintiff to it.” Id. at 900 (emphasis added); see also Wood, 879 F.2d at 588.

Our subsequent cases have further demarcated the outer bounds of the state-created danger doctrine. These cases have only highlighted the requirement that, at a minimum, a due process claim must be based on an affirmative exercise of state power that creates a risk which, but for the state’s affir- mative action, would not have existed. For instance, in Penilla v. City of Huntington Park, 115 F.3d 707 (9th Cir. 1997), we found a due process violation where police officers responded to a 911 call, “examined [the plaintiff], found him to be in 2258 KENNEDY v. RIDGEFIELD grave need of medical care, canceled the request for parame- dics, broke the lock and door jamb on the front door of [the plaintiff’s] residence, moved him inside the house, locked the door, and left.” Id. at 708. Under these circumstances, we determined that the state created a danger to the plaintiff which, but for its affirmative unlawful acts, would not have existed.3 Likewise, in Munger, we found a cognizable due process violation where police officers ejected the plaintiff from a bar late at night when the outside temperatures were subfreezing. 227 F.3d at 1087. Although the officers knew 3 The majority cites to Penilla for the proposition that this Court has “specifically rejected the ‘danger creation’ versus ‘danger enhancement’ distinction.” Maj. Op. at 2243 n.4; Penilla, 115 F.3d at 710 (“The critical distinction is not , as appellants allege, an indeterminate line between dan- ger creation and enhancement, but rather the stark one between state action and inaction in placing an individual at risk.”). This reading of Penilla is misguided. Reading the larger passage in which this sentence appears produces a different picture: The officers argue that under DeShaney, a constitutional duty to provide care is only triggered when a person is in custody. We reject this argument. . . . We have interpreted DeShaney to mean that if affirmative con- duct on the part of a state actor places a plaintiff in danger, and the officer acts in deliberate indifference to that plaintiff’s safety, a claim arises under § 1983. In Grubbs we explained: DeShaney did not rule that custody was required where the state affirmatively causes the harm . . . . DeShaney thus sug- gests that had the state created the danger, [plaintiff] might have recovered even though he was not in custody. The critical distinction is not, as appellants allege, an indetermi- nate line between danger creation and enhancement, but rather the stark one between state action and inaction in placing an indi- vidual at risk. Id. at 710 (citations omitted). Our opinion in Penilla focused on the new danger that the officers created for Penilla: that by affirmatively calling off the paramedics and moving him from his porch—where neighbors and a passerby had seen his predicament and rendered aid—into his locked house, police isolated Penilla, making it impossible for him to receive medical care. KENNEDY v. RIDGEFIELD 2259 that the plaintiff was intoxicated and was wearing only a t- shirt and jeans, they prevented him from driving his truck or reentering the bar. Id. at 1084-85. Presented with these facts, we held that the state affirmatively acted to place the plaintiff in danger that would not have existed without state action. Id. at 1087.

In those cases where we have declined to find a cognizable due process violation, we have generally emphasized the unforeseeable nature of the plaintiff’s injuries, that the danger facing the plaintiff existed independent of state action, or the absence of the requisite mental state. For instance, in Huffman v. County of Los Angeles, 147 F.3d 1054, 1061 (9th Cir. 1998), we declined to find municipal liability under § 1983 where the plaintiff was shot during a barroom brawl with an off-duty deputy employed by the Los Angeles County Sher- iff’s Department. Finding that the risk to the plaintiff was an unforeseeable consequence of a county policy requiring off- duty officers to carry a firearm, we held that “the danger- creation plaintiff must demonstrate, at the very least, that the state acted affirmatively, and with deliberate indifference, in creating a foreseeable danger to the plaintiff, leading to the deprivation of the plaintiff’s constitutional rights.” Id. (cita- tions omitted); see also Lawrence v. United States, 340 F.3d 952, 957 (9th Cir. 2003) (citing Penilla and Munger, and observing that “in each of the cases in which we have applied the danger-creation exception, ultimate injury to the plaintiff was foreseeable”). Similarly, in Lawrence, 340 F.3d at 954, we declined to find a Fifth Amendment violation in a Bivens action where a juvenile plaintiff alleged that she was sexually abused by a convicted drug offender participating in the Fed- eral Witness Security Program; the plaintiff alleged that the offender could not have obtained employment at a group home where she was a resident but for the assistance of fed- eral officers.4 Although we found it foreseeable that a con- 4 There may be some latent dispute regarding whether the “proximate cause” requirement noted in Huffman, 147 F.3d at 1061, and Lawrence, 2260 KENNEDY v. RIDGEFIELD victed drug offender might attempt to distribute illegal drugs to children with whom he came into contact, we found the plaintiff’s injuries an unforeseeable consequence of the offi- cial action. Id. at 957.

Finally, in Nicholas v. Wallenstein, 266 F.3d 1083 (9th Cir. 2001), the most factually similar case in our case law, county jail employees brought suit against the jail commander after he publicly disclosed the identities of employees who had been involved in the restraint and removal of a deceased pris- oner. Upon learning their identities, the deceased prisoner’s family and friends harassed and assaulted the employees. Id. at 1085-86. Citing Wood, the employees contended that the state had acted with deliberate indifference because their supervisors did not promptly notify them of the release of their identities and did not take steps to protect them from the dangers that ultimately became apparent. Id. at 1087. We ruled in favor of the state, finding that the plaintiffs had not established that the commander acted with deliberate indiffer- ence to known or obvious dangers, even though he knew when he released the records that the deceased prisoner’s family and friends believed that personnel connected with the jail were responsible for his death. Id. In doing so, we rea- soned that the jail authorities could not have reasonably con- cluded that the prisoner’s family and friends would be likely to engage in open violence. Id. (“Knowing that the crowd was angry was not knowing that they would take criminal mea- sures to make the jailors or their health helpers pay.”).

340 F.3d at 957, is in addition to, or a mere rephrasing of, the requirement that the danger to the plaintiff must have been “known or obvious” and the state actor must have acted with deliberate indifference to the danger. See, e.g., Grubbs II, 92 F.3d at 899-900. Nonetheless, for purposes of the instant case, the relevance of Huffman, Lawrence and Wallenstein derives simply from their recognition that traditional causation principles are not wholly suspended in the context of a constitutional tort suit premised on state-created danger. KENNEDY v. RIDGEFIELD 2261 2. Factors for Analysis

As our cases illustrate, we typically consider a number of factors in determining whether the plaintiff has successfully stated a due process violation: (1) whether the act was directed toward a specific plaintiff or the public at large, see, e.g., Wood, 879 F.2d at 590 (reasoning that the state’s action “distinguish[ed] [the plaintiff] from the general public and trigger[ed] a duty of the police to afford her some measure of peace and safety”); cf. Huffman, 147 F.3d at 1061 & n.4 (sug- gesting, but not deciding, that a plaintiff must show that “the danger created by a state official is directed toward a particu- lar plaintiff, as opposed to being directed toward the general public”); (2) whether the government acted affirmatively or simply failed to act, see, e.g., Grubbs I, 974 F.2d at 121 (requiring “affirmative conduct on the part of the state in plac- ing the plaintiff in danger”); Munger, 227 F.3d at 1086 (phrasing the inquiry as “whether [the state] did in fact affir- matively place [the plaintiff] in danger”); (3) whether the gov- ernment’s act caused the harm, see, e.g., Grubbs I, 974 F.2d at 121 (finding state-created danger where the state’s action “create[d] an opportunity for [the inmate] to assault [the plaintiff] that would not otherwise have existed” (emphasis added)); Penilla, 115 F.3d at 710 (same); Munger, 227 F.3d at 1087 (same); Ketchum v. Alameda County, 811 F.2d 1243, 1247 (9th Cir. 1987) (“[T]here is no constitutional right to be protected by the state against being murdered by criminals or madmen. It is monstrous if the state fails to protect its resi- dents against such predators but it does not violate the due process clause of the Fourteenth Amendment.”) (quoting Bowers v. DeVito, 686 F.2d 616, 618 (7th Cir.1982)); and (4) whether the government acted with the requisite culpability, see, e.g., Grubbs II, 92 F.3d at 900 (requiring the plaintiff to show that the state official “acted with deliberate indifference to the known or obvious danger” (emphasis added)); Wallen- stein, 266 F.3d at 1087 (same); Penilla 115 F.3d at 710 (same). Cf. Armijo ex rel. Chavez v. Wagon Mound Pub. Schs., 159 F.3d 1253, 1264 (10th Cir. 1998) (adding a fifth 2262 KENNEDY v. RIDGEFIELD factor which considers whether the government completely removed all of the plaintiff’s protection); Russell v. Gregoire, 124 F.3d 1079, 1093 n.10 (9th Cir. 1997) (stating, in dicta, that “a state has no general duty to protect individuals against potential harm by third parties unless the state creates the dan- ger and removes the individual’s ability to protect himself” (citations omitted)). These factors closely parallel those used by other circuits recognizing the doctrine. See, e.g., Uhlrig v. Harder, 64 F.3d 567, 574 (10th Cir. 1995) (requiring the plaintiff to show that (1) he “was a member of a limited and specifically definable group; (2) Defendants’ conduct put [him] and the other members of that group at substantial risk of serious, immediate and proximate harm; (3) the risk was obvious or known; (4) Defendants acted recklessly in con- scious disregard of that risk; and (5) such conduct, when viewed in total, is conscience shocking”). My disagreement with the majority’s findings and conclusions centers on the second, third and fourth factors.

The Supreme Court has yet to recognize the state-created danger doctrine, and the circuit courts have yet to construct a unified approach either to the state-created danger inquiry or to the role that causation principles should play in the analy- sis. However, each court recognizing the theory has required, at a minimum, a showing that the government’s act was the “but-for cause” that put the plaintiff in a position of danger she would not otherwise have faced. See, e.g., Carlton v. Cle- burne County, 93 F.3d 505, 508 (8th Cir. 1996) (collecting cases and noting that in each case where a cognizable due process violation was found “the individuals would not have been in harm’s way but for the government’s affirmative actions”); Reed v. Gardner, 986 F.2d 1122, 1125 (7th Cir. 1993) (finding the evidence sufficient to support summary judgment for police officers where “without state interven- tion, the same danger would exist”); Salas v. Carpenter, 980 F.2d 299, 309-10 (5th Cir. 1992) (holding a city not liable for declining assistance from a SWAT team and taking a hard line with a hostage taker); Jackson v. City of Joliet, 715 F.2d KENNEDY v. RIDGEFIELD 2263 1200, 1204-05 (7th Cir. 1983) (holding officers not liable because they “did not create but merely failed to avert dan- ger” by not rescuing victims from a burning car more promptly). We have never recognized a state-created danger where the state was merely a “proximate cause” rather than the cause-in-fact of the plaintiff’s injuries. We have not imported common law tort principles to this doctrine. As the Supreme Court observed in DeShaney:

It may well be that, by voluntarily undertaking to protect [the plaintiff] against a danger it concededly played no part in creating, the State acquired a duty under state tort law to provide him with adequate protection against that danger. . . . But the claim here is based on the Due Process Clause of the Fourteenth Amendment, which, as we have said many times, does not transform every tort committed by a state actor into a constitutional violation.

489 U.S. at 201-02 (citations omitted). In short, our cases, as well as those of our sister circuits, demand that the state’s affirmative act must, at the very least, be the cause-in-fact of the plaintiff’s injury.

My motive for further belaboring the federal reports with a dissent stems primarily from my conviction that Kennedy has not alleged facts sufficient to support a due process viola- tion; her case against Officer Shields sounds in negligence, albeit negligence with tragic consequences. The majority has run afoul of our own cases and the Court’s caution in DeShaney. I address these issues more fully below.

B. Constitutional Inquiry

The majority finds fault with two of Officer Shields’s actions: (1) notifying Angela Burns of Kennedy’s allegations prior to informing Kennedy that he was about to do so;5 and 5 The majority makes some effort to suggest that their theory of this case does not turn on the question of whether officer Shields contacted Ken- 2264 KENNEDY v. RIDGEFIELD (2) promising to increase police surveillance on the night of the shooting. Maj. Op. at 2245-46. Neither of these, consid- ered independently or together, will support a due process violation.

1. Notifying Burns Prior to Informing Kennedy

The majority concludes that Officer Shields “created an opportunity for Burns to assault the Kennedys that otherwise would not have existed.” Id. at 2242 (quotations omitted). Kennedy has not addressed how much advance warning she desired, nor whether she effectively communicated the extent of warning she desired to Shields; however, she insists that she made it clear that she wanted to be notified before the Burnses were informed of her allegations. The majority asserts that, had Kennedy received prior warning, she and her family would have had the opportunity to take additional pre- cautions. Id. at 2242. The majority reaches this conclusion despite the fact that Shields warned her within fifteen minutes of his discussion with Angela Burns and that the Kennedys subsequently made a conscious choice to remain in their home for the evening. The majority finds this flipflop of no more than fifteen minutes to be of constitutional magnitude. As the Kennedys were shot many hours later, I do not see how receiving warning fifteen minutes earlier would have made any difference whatsoever. Nonetheless, in light of the

nedy before or after he spoke to Burns. See Maj. Op. at 2242 (“The exis- tence of this danger does not depend . . . on a difference of fifteen-minutes . . . .”). However, if this is the case, it becomes entirely unclear precisely what Shields’s misconduct was, and accordingly even more difficult for Shields to have known that his conduct was not merely wrong, but that it violated Kennedy’s constitutional rights. Moreover, Kennedy only argues that Shields’s error was telling her before he told Burns. Thus, the opin- ion’s vague contrary language notwithstanding, the majority opinion must rest on the fact that Shields informed Burns before telling Kennedy he was going to do so. See Maj. Op at 2242 (“[Shields] did [not] give Kennedy a reasonable opportunity to protect her family . . . .”). I therefore treat it as such. KENNEDY v. RIDGEFIELD 2265 information Kennedy communicated to Officer Shields regarding Michael’s past misbehavior, the majority holds that “Shields’s actions both created and aggravated the risk Plain- tiff faced from Burns” Id. at 2250.

There is nothing in the record to support the claim that Shields increased the risk facing the Kennedy family by noti- fying Angela Burns of the allegations. Notifying Michael Burns was an inevitable consequence of Kennedy’s allega- tions of child molestation; at some point either the police or CAIC was going to have to talk with Burns about the allega- tions.6 Kim Kennedy was anxious because she knew that Michael Burns would have to be informed, and she feared what he might do when he was. It was this fear that motivated her to contact police at least six times to inquire whether the Burnses had been contacted yet. In none of these numerous phone calls did she try to dissuade the authorities from ever contacting Burns; she knew that it was only a matter of time. The dilemma for her was whether she would know when Burns was contacted, and would therefore be able to take pre- cautions. To that end, she made every effort to ensure that she would be notified when Burns was made aware of these charges.

Prior to the shooting, Kennedy’s only direct contact with law enforcement officials was with Officers Shields and 6 In my view, whether or not Burns would inevitably discover the allega- tions against him is not, as the majority states, “beside the point”; it is a question of crucial importance for this case. Maj. Op. at 2242 n.3. If this was a specific danger from Burns that Kennedy had to face, it becomes clear that Shields could not have created it. Moreover, the majority’s suggestion that Burns might never need to be notified of the allegations against him strains credulity. Id. The majority’s quotations from the record only suggest that Burns should have been noti- fied at the end of the investigation. Moreover, these same quotations also establish that this was done so that an officer questioning Burns would be better able to identify whether he was lying, not because it reduced the chance of a violent response. 2266 KENNEDY v. RIDGEFIELD Doriot of the RPD. However, pursuant to an inter-local agree- ment, the task of investigating Kennedy’s molestation com- plaint was performed solely by a separate law enforcement unit, the Child Abuse Intervention Center (“CAIC”). So far as Shields knew, Kennedy had had no contact with CAIC and was relying on conversations with him and Officer Doriot to monitor the case. Shields had no authority over CAIC, and therefore had no way of ensuring that Kennedy received noti- fication before CAIC made contact with the Burns family regarding her allegations. Indeed, from Shields’s perspective, he represented Kennedy’s best chance of receiving timely notification of any contact with the Burnses. Judging from Kennedy’s repeated calls to Shields, Kennedy took a similar view.

The majority’s statement that “[o]f all the possible actions [Shields] could take, . . . he took the one most feared by Ken- nedy” is simply false. Maj. Op. at 2245. The scenario Ken- nedy most feared was that Burns would become aware of the allegations and she would not know, and therefore would not be able to take appropriate precautions. Thus, when Shields decided to inform Burns of the allegations himself, he was ensuring that Kennedy was spared the possibility she feared most—that Burns would be notified and she would be unaware. And, by Kennedy’s own testimony, Officer Shields informed her immediately after contact was made, at approxi- mately 4:30 in the afternoon.

The majority attempts to shoehorn Shields’s behavior in this case into the mold of the supervisor in Grubbs I. This is an exceedingly poor analogy. The supervisor in Grubbs I cre- ated the danger to the detention center nurse by essentially ordering her to work alone with a known violent sex offender. If he had not done so, the nurse would presumably never have been alone with the offender, and would therefore not have been in any danger from him. Here, Burns would have to be informed eventually; the only question was whether Kennedy would know that he had been informed. Nor did Shields facil- KENNEDY v. RIDGEFIELD 2267 itate Michael Burns’s access to Kennedy. Unlike the nurse in Grubbs, Kennedy was well aware that she was already exposed to a very real danger, and that this danger existed apart from any action or conduct by Officer Shields. Rather than increasing the risk facing the Kennedy family, Shields’s prompt notification appears to have given Kennedy her best chance for escape.

Yet, even if Officer Shields had increased the risk facing the plaintiff, this would not constitute a due process violation. See, e.g., Huffman, 147 F.3d at 1061 (“The danger-creation exception to DeShaney does not create a broad rule that makes state officials liable under the Fourteenth Amendment whenever they increase the risk of some harm to members of the public.”). The City of Ridgefield did not create Michael Burns’s violent reaction any more than Winnebago County created the violent beatings that resulted in brain damage to Joshua DeShaney. See DeShaney, 489 U.S. at 193. The major- ity’s holding impermissibly circumvents DeShaney by rede- fining the cause of action as one premised on a “state-created danger.” I therefore cannot support the majority’s holding that, like the supervisor in Grubbs I, Shields created “an opportunity for Burns to assault the Kennedys that otherwise would not have existed.” Maj. Op. at 2241-42 (emphasis added) (quotations omitted); Grubbs I, 974 F.2d at 121.

Nor can Shields’s conduct be characterized as manifesting “deliberate indifference” to the dangers faced by the Kenne- dys. As the majority acknowledges, “the standard in this cir- cuit [is] not gross negligence but ‘deliberate indifference to a known, or so obvious as to imply knowledge of, danger.’ ” Maj. Op. at 2245; see Grubbs II, 92 F.3d at 898; see also DeShaney, 489 U.S. at 201-02. Grubbs II’s deliberate indif- ference standard requires a showing that the “ ‘defendant rec- ognizes the unreasonable risk and actually intends to expose the plaintiff to such risks without regard to the consequences to the plaintiff.’ ” Grubbs II, 92 F.3d at 899 (quoting Uhlrig, 64 F.3d at 573 n.8). Phrased another way, the defendant must 2268 KENNEDY v. RIDGEFIELD “have actual knowledge of, or willfully ignore, impending harm,” meaning “the defendant knows that something is going to happen but ignores the risk and exposes someone to it.” Id. at 900 (emphasis in original).7

Even if Officer Shields knew of Michael Burns’s propensities—the allegations that he had threatened a class- mate, tortured a cat, and assaulted his girlfriend—Shields could not have anticipated as an “obvious consequence” that Michael would enter the Kennedys’ home and murder Jay and assault Kim. See Maj. Op. at 2243; Wallenstein, 266 F.3d at 1087. Although his previous misconduct included disturbing juvenile violence, nothing in his record should have made it obvious that the thirteen-year-old Burns might attempt to murder members of the Kennedy family with a firearm. Indeed, the record suggests that both Shields and the Kenne- dys failed to appreciate the extent of the danger that Michael posed. Under these circumstances, it cannot be said that Offi- cer Shields had “actual knowledge of, or willfully ignore[d], impending harm.” Grubbs II, 92 F.3d at 900; see also Wallen- stein, 266 F.3d at 1087 (“[It] has not been shown . . . that . . . [friends and family of the deceased prisoner] would have the 7 The majority opinion incorrectly characterizes my position as “requir- ing foreseeability of the specific injury Burns in fact inflicted on the Ken- nedys.” Maj. Op. at 2244 n.5. I agree with the majority that “the exact injury inflicted by a third party” need not have been foreseeable. Id. How- ever, Kennedy argues here that Shields’s misconduct was informing Michael Burns that her daughter had made allegations against him without giving her prior warning. By Kennedy’s own testimony, Shields made her aware that Burns had been notified immediately after he had notified Burns. It certainly was not foreseeable that this difference—telling Ken- nedy immediately before or immediately after informing Burns—would lead to the type of injuries that she suffered. See also Wallenstein, 266 F.3d at 1087 (finding that it was not foreseeable that the friends and family of a deceased inmate who blamed jail personnel for the death “would take criminal measures” against said jail personnel); id. (“The most serious incident, assault with a gun, was the sort of opportunistic crime which could not have easily been anticipated nor easily guarded against.”). KENNEDY v. RIDGEFIELD 2269 capacity and sustained desire to wreak vengeance on the offi- cers and nurses involved [with his death].”).

Even assuming, arguendo, that Shields recognized the risk that Kennedy faced from Burns, his actions can hardly be said to demonstrate “deliberate indifference” to it. Even if Shields’s actions were misguided in hindsight—and it is not clear that any other reasonable officer would not have done the same thing—all the evidence suggests that he was moti- vated by a desire to ensure that Kennedy would know exactly when Burns became aware of her daughter’s allegations. There is simply no evidence that Shields acted with deliberate indifference to any known or obvious risks Kennedy faced.8 Without the requisite mental state, there can be no constitu- tional violation premised on state-created danger. See, e.g., Grubbs II, 92 F.3d at 898; Wood, 879 F.2d at 588. I would hold that Kennedy failed to state a constitutional violation arising from the prompt notification that she received regard- ing Shields’s contact with Angela Burns.

2. Promising Police Surveillance

The majority correctly recognizes that officer Shields’s assurances of a police patrol on the evening of the shooting do not provide an independent basis for a due process viola- tion. Maj. Op. at 2243 (“[W]e do not rest our judgment that Shields affirmatively created a danger on that assurance . . . .”). However, I cannot agree with the majority’s conten- tion that, by assuring Kennedy “that the police would patrol the area,” Shields somehow aggravated the risks that Kennedy 8 Considering the alternative courses of conduct Shields could have taken to escape liability under the majority’s theory only highlights the artificiality of the majority’s analysis. Under the majority’s theory, Shields could simply have reversed the order in which he visited the residences of the plaintiff and her would-be assailant, or called Kim Kennedy on his cell phone from the Burnses’ doorstep. I cannot agree with the majority’s posi- tion that this flipflop of no more than fifteen minutes is of constitutional magnitude. 2270 KENNEDY v. RIDGEFIELD faced. Id. at 2233; id. at 2243 (“Instead, [it] was an additional and aggravating factor, making [Kennedy] more vulnerable to the danger he had already created [by notifying Burns of the allegations against him before telling Kennedy that he was about to do so].”). Kennedy does not claim that the RPD failed to patrol the area on the evening of the shooting, nor does she allege that Officer Shields made any false claims to her about the efficacy of police patrols in providing protection in similar cases. I do not see how Officer Shields’s statement that the police would patrol the area made the Kennedys “more vulnerable.” See DeShaney, 489 U.S. 189 (finding mul- tiple attempted but failed interventions by social services insufficient to create a due process violation); Balistreri v. Pacifica Police Dep’t, 901 F.2d 696, 700 (9th Cir. 1990) (cit- ing DeShaney and declining to find a due process violation where the plaintiff’s allegations amounted to the assertion that “state actors knew of her plight and affirmatively committed to protect her”).

The majority attempts to justify its statement by analogiz- ing the facts of this case to those of Grubbs I. This compari- son does not help the majority’s case. In Grubbs I, we relied on the state’s misrepresentation merely as a means for bolster- ing our conclusion that the state’s affirmative act of directly placing the plaintiff in a dangerous situation—namely, assign- ing her to work alone with a known violent sex offender— created a risk that would not otherwise have existed. See Grubbs I, 974 F.2d at 121; see also Munger, 227 F.3d at 1086 (noting that the court in a state-created danger case “must determine whether [the state] did in fact affirmatively place [the plaintiff] in danger”). Here, Kennedy does not allege that the government lied about the risks she would face, but rather that she relied on government protective measures which failed her. While it is undeniably tragic that police patrols were unsuccessful in preventing Burns’s attack, this is cate- gorically different from Grubbs I, where the government actively misrepresented the risks facing the plaintiff. I there- KENNEDY v. RIDGEFIELD 2271 fore believe the majority’s reasoning on this issue to be flawed.

In sum, I would hold that Kennedy failed to establish a due process violation arising from Officer Shields’s actions either in notifying Michael Burns of her allegations prior to warning her, or in offering to increase surveillance on the evening of the shooting. Accordingly, I would hold that she failed to establish a cognizable due process violation premised on state-created danger.

C. Qualified Immunity Inquiry

Even assuming that Kennedy has established a due process violation premised on state-created danger, in order to bind this case over for trial we must determine that the constitu- tional right at issue was “clearly established” at the time of the events in question. We must hold that a “reasonable offi- cial” in Officer Shields’s position “would understand that what he is doing violates that right,” Saucier, 533 U.S. at 202, keeping in mind that “officials will not be liable for mere mis- takes in judgment, whether the mistake is one of fact or one of law.” Butz v. Economou, 438 U.S. 478, 507 (1978). Indeed, “[e]ven defendants who violate constitutional rights enjoy a qualified immunity that protects them from liability for dam- ages unless it is further demonstrated that their conduct was unreasonable under the applicable standard.” Davis v. Scherer, 468 U.S. 183, 190 (1984). As the Court has repeat- edly emphasized, “the qualified immunity defense . . . pro- vides ample protection to all but the plainly incompetent or those who knowingly violate the law.” Malley v. Briggs, 475 U.S. 335, 341 (1986); see also Burns v. Reed, 500 U.S. 478, 494-95 (1991). Particularly in a context where the potential for liability may chill lawful and socially desirable behavior at the edge of the “forbidden zone,” qualified immunity ensures that “officials can act without fear of harassing litiga- tion” and “can anticipate when their conduct may give rise to liability for damages.” Davis, 468 U.S. at 195. 2272 KENNEDY v. RIDGEFIELD Imbued with notions of “reasonableness” and “fair warn- ing,” the “concern of the immunity inquiry is to acknowledge that reasonable mistakes can be made as to the legal con- straints on particular [official] conduct.” Saucier, 533 U.S. at 205. The central dispositive inquiry essential to finding a right “clearly established” is “whether it would be clear to a reason- able officer that his conduct was unlawful in the situation he confronted.” Id. at 202. Importantly, our analysis must acknowledge and evaluate the specific context of the situation confronted by the official. Id.; see also Brosseau v. Haugen, ___ U.S. ___, ___, 125 S.Ct. 596, 599 (2004) (“It is important to emphasize that this inquiry ‘must be undertaken in light of the specific context of the case, not as a broad general propo- sition.’ ” (quoting Saucier, 533 U.S. at 201)). With these instructions in mind, I have no hesitation in concluding that Officer Shields is entitled to qualified immunity.

The majority holds that Officer Shields’s behavior violated Kennedy’s clearly established constitutional rights because it finds the case “not ‘meaningfully distinguishable’ from Grub- bs.” Maj. Op. at 2248. I disagree. Grubbs I does not even begin the heavy lifting necessary to sustain the majority’s conclusions.

The majority writes that “[i]n Grubbs, as in this case, a state official affirmatively acted: supervisor Grubbs assigned a violent sex offender to work closely with [the nurse], and Officer Shields notified Burns, leaving Kennedy unable to protect her family.” Id. at 2249. Indeed, Shields did take an affirmative act. However, the danger in Grubbs—being alone with a known violent sex offender—was entirely avoidable, while the danger in this case—that Burns might react vio- lently when he discovered the allegations against him—was not within Shields’s control. The only danger that Shields was able to ameliorate was the possibility that Kennedy would not be aware that Burns had learned of the allegations against him; Shields did, in fact, prevent this scenario. Moreover, in Grubbs, the supervisor made false representations so that the KENNEDY v. RIDGEFIELD 2273 victim could not evaluate her level of danger and take appro- priate precautions. Here, Shields made no misrepresentations and Kennedy already knew the risks. I therefore find the majority’s statement that, “At bottom Kennedy’s claim is exactly like [the nurse in Grubbs], i.e., that a state actor ‘enhanced [her] vulnerability to attack by misrepresenting to her the risks’ she faced” mystifying. Id. at 2249.

The majority further likens this case to Grubbs because Shields’s action “made plaintiffs vulnerable to a particular- ized danger they would not have faced but for that action.” Id. at 2249 (emphasis added); see also id. (“[I]n this case, as in Grubbs, Shields used his ‘authority as a state . . . officer to create an opportunity for [Burns] to assault [Kennedy] that would not have otherwise existed.’ ”) (emphasis added) (alter- ations and omission in original). The risk that Burns would react violently when he discovered the allegations Kennedy had made against him existed entirely apart from any action attributable to Officer Shields. In fact, the risk to Kennedy would have been even greater if Kennedy was unaware that Burns had learned of the allegations.

In short, I cannot join the majority’s holding that Grubbs I put Officer Shields on notice that by responding to Kenne- dy’s phone message, informing Angela Burns of Kennedy’s allegations, immediately notifying Kennedy of as much, and offering to increase surveillance in the neighborhood, he was violating her Fourteenth Amendment due process rights—and that the violation was so obvious that Shields should have known it.

No case of which I am aware, either in our circuit or any other, has found a cognizable due process violation on facts remotely analogous to these.9 On the contrary, the closest case (Text continued on page 2275)

9 This includes all of the cases to which the majority cites to stress the prevalence of the state-created danger doctrine, see Maj. Op. at 2239 n.1, none of which give notice here. Some of these cases involve facts such as 2274 KENNEDY v. RIDGEFIELD those in Wood, where the police needlessly left people that were in some way helpless in a dangerous environment; these cases are inapplicable. See Kneipp v. Tedder, 95 F.3d 1199 (3d Cir. 1996) (holding that allegations that police left heavily intoxicated pedestrian alone to walk home on cold night could establish violation); Reed v. Gardner, 986 F.2d 1122 (7th Cir. 1993) (denying summary judgment for defendants where police arrested a driver and left an intoxicated passenger in the vehicle with the keys); White v. Rochford, 592 F.2d 381, 384 (7th Cir. 1979) (finding that the complaint alleged a violation where police “left helpless minor children subject to inclement weather and great physical danger without any appar- ent justification”). Others involve police who were aware of a clear and obvious danger and actively chose not to provide their usual level of pro- tection out of a desire to encourage the would-be violent actors. See Dwares v. City of New York, 985 F.2d 94, 99 (2d Cir. 1993) (finding a possible violation where complaint alleged “that the officers conspired with . . . ‘skinheads’ to permit [them] to beat up flag burners with relative impunity, assuring [them] that . . . they would not be impeded or arrested,” thereby purposefully “increas[ing] the likelihood that [they] would assault demonstrators”); Freeman v. Ferguson, 911 F.2d 52 (8th Cir. 1990) (hold- ing that complaint alleging that police chief failed to act to perform his duties because of his close personal relationship with perpetrator was insufficient to establish a violation, but that specific allegations that he actively prevented other officers from doing so could support a claim). With a single exception, all of the cases cited by the defense which have facts that are even arguably close to those presented in this case were decided in favor of the government actors. See Butera v. District of Columbia, 235 F.3d 637, 652 (D.C. Cir. 2001) (dismissing suit by estate of slain undercover informant because the informant’s “constitutional right to protection by the District of Columbia from third-party violence was not clearly established”); Uhlrig v. Harder, 64 F.3d 567 (10th Cir. 1995) (dismissing suit alleging that state created the danger that led to a thera- pist’s death by eliminating mental hospital’s special unit for criminally insane); Wells v. Walker, 852 F.2d 368 (8th Cir. 1988) (finding no viola- tion, and, at most, negligence, when police released a convict from cus- tody outside of a store without warning the owner that he was dangerous, and the convict killed the owner); Bowers v. DeVito, 686 F.2d 616, (7th Cir. 1982) (finding that no right had been violated when inmate who had repeatedly attacked women with knives was released from commitment and subsequently killed a woman with a knife). The only case finding a violation which is at all similar is Kallstrom v. City of Columbus, 136 F.3d 1055 (6th Cir. 1998), in which the court found that undercover police offi- KENNEDY v. RIDGEFIELD 2275 to this one in our circuit concluded that the plaintiff could not establish a due process violation. As my earlier discussion of Nicholas v. Wallenstein makes clear, its facts are strikingly similar: A state officer released incident reports with the plaintiff prison workers’ identities to the angry family and friends of a deceased prisoner; plaintiffs were immediately harassed and assaulted. 266 F.3d at 1084-86. When the state official “released the incident reports he knew that the crowd to whom he was releasing them believed that personnel con- nected with the jail had killed” the deceased prisoner and that the reports “would excite the crowd.” Id. at 1087. Yet, when presented with these facts a few years ago, we held that the “plaintiffs failed to produce evidence that would create a tri- able issue of material fact showing that the danger was known or obvious to the defendants.” Id. at 1085. What we stated there bears repeating here: “Knowing that the crowd was angry was not knowing that they would take criminal mea- sures to make the jailors or their health helpers pay.” Id. at 1087.

We have always drawn a sharp distinction between facts demonstrating that police action created the danger to the per-

cers who had worked to convict members of a gang with propensity for violence and intimidation were entitled to an injunction requiring the city to provide them with notice before publicly releasing certain information from their personnel files that the city had assured them would be kept confidential. This information included “the officers’ addresses and phone numbers; the names, addresses, and phone numbers of immediate family members; the names and addresses of personal references; the officers’ banking institutions and corresponding account information, including account balances; their social security numbers; responses to questions regarding their personal life asked during the course of polygraph exami- nations; and copies of their drivers’ licenses, including pictures and home addresses.” Id. at 1059. Nonetheless, this case is easily distinguishable because while in Kennedy’s case disclosure was inevitable, the gang members in Kallstrom would never have learned this information absent a disclosure by the city. Moreover, the majority does not rely on this case in its opinion. 2276 KENNEDY v. RIDGEFIELD son and facts demonstrating a danger that existed without police action.10 See Wood, 879 F.2d at 589-90. In addition, since Grubbs II, we have required plaintiffs to meet a strin- gent culpability requirement designed to prevent the imposi- tion of § 1983 liability for negligent conduct, even grossly negligent conduct. 92 F.3d at 899-900. And since Huffman, 147 F.3d at 1061, and Lawrence, 340 F.3d at 957, we have emphasized that the requisite culpability must relate to conse- quences which were foreseeable. The majority’s conclusion in this case does not simply whittle away at these requirements; it completely reinvents them and then declares them “clearly established.”11

I cannot envision how it “would be clear to a reasonable officer that his conduct was unlawful” in the situation at issue in this case. Saucier, 533 U.S. at 202. Assuming, arguendo, that Kennedy’s allegations are sufficient to state a constitu- tional violation, there is no way Shields could have antici- pated that his fifteen-minute delay in notifying Kennedy, combined with his statement that he would patrol the area, was depriving her of her rights under the Due Process Clause of the Fourteenth Amendment. Even if he had read Grubbs I —but especially if he had read Wallenstein—Officer Shields 10 The majority relies on a single sentence in Penilla to suggest other- wise; this reading is flawed. See footnote 3, supra. 11 The majority claims that I have improperly engaged in “an elaborate fact-matching exercise” to demonstrate that “none of our state-created danger cases clearly enough established the requisite notice,” and that this exercise is “misguided” and “analytically flawed.” Maj. Op. at 2247 n.7. While I wholeheartedly agree with the majority that an “exact factual predicate” is not required for a right to be clearly established, neither do I believe that we should decide whether a right is clearly established with- out considering the facts of the other cases in which we have considered that right. I believe the majority’s unwarranted extension of the law makes a mockery of prior decisions emphasizing the importance of providing fair warning to government officials. I think this approach is unwise generally, but that it is especially troubling here, where the case with the closest facts is clearly not Grubbs I, but Wallenstein—a case where we ruled in favor of the government official. KENNEDY v. RIDGEFIELD 2277 could not have known that his conduct would violate “clearly established” constitutional rights. See Meyers v. Redwood City, 400 F.3d 765, 774 (9th Cir. 2005) (“Even with a copy of Harris in their back pockets, the officers could not have determined at what point in the middle of this messy reposses- sion they deprived Meyers of her property without due pro- cess of law.”). I would hold that, taking into account the “specific context of th[is] case,” the right was not clearly established at the time Officer Shields acted, and Shields is thus entitled to qualified immunity. Saucier, 533 U.S. at 201.

III. CONCLUSION

Given the tragic circumstances in which this case arises, the Court’s instruction in DeShaney seems especially apt: “Judges and lawyers, like other humans, are moved by natural sympa- thy in a case like this” to find a way for Kennedy and her fam- ily “to receive adequate compensation for the grievous harm inflicted upon them. But before yielding to that impulse, it is well to remember once again that the harm was inflicted not by the State,” but by Michael Burns. 489 U.S. at 202-03. The people of Washington may prefer, and are free to adopt, a sys- tem of tort liability which would place upon the State and its officials the responsibility for situations such as the present one. “But they should not have it thrust upon them by this [c]ourt’s expansion of the Due Process Clause of the Four- teenth Amendment.” Id. at 203.

I respectfully dissent.