RISVIN DE LEON LOPEZ V. MERRICK GARLAND

20-71529 | 9th Cir. | Oct 21, 2022

FILED OCT 21 2022

MOLLY C. DWYER, CLERK U.S. COURT OF APPEALS FOR PUBLICATION UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT RISVIN VALDEMAR DE LEON No. 20-71529 LOPEZ,

Agency No. Petitioner, A200-569-789 v. OPINION MERRICK B. GARLAND, Attorney General,

Respondent. On Petition for Review of an Order of the Board of Immigration Appeals Argued and Submitted September 13, 2021 Pasadena, California Before: Ronald M. Gould, Marsha S. Berzon, and Daniel P. Collins, Circuit Judges. Opinion by Judge Berzon; Dissent by Judge Collins

SUMMARY [*] Immigration

Granting Risvin Valdemar De Leon Lopez’s petition for review of a decision of the Board of Immigration Appeals upholding the denial of protection under the Convention Against Torture, and remanding, the panel concluded that the record in this case compelled the conclusion that two of De Leon’s attackers during his first attack were police officers, that the police officers’ participation in the incident showed acquiescence on the part of the Guatemalan government, and that the agency disregarded several important circumstances in concluding that De Leon would not likely be subjected to future torture.

The agency found that De Leon’s belief that some of his attackers during his first incident were police officers was based on speculation, and that the two individuals wearing police uniforms did so only to make themselves look like police officers. The panel concluded that the agency’s assessment was not based on substantial evidence. The panel explained that De Leon presented four significant pieces of evidence showing that the two individuals were police officers: first, De Leon had known the two men for about twenty years before the incident, and had been told by people in his small town that they were in fact police officers; second, the two individuals were wearing police uniforms at the time of the attack; third, they were armed

* This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader. with visible handguns that De Leon recognized as of the kind, and in the holstered position, typical of national police officers; and fourth, the Public Ministry to which De Leon reported the attack indicated that the two individuals had left the police force shortly after the attack, thus confirming that they were members of the police during the incident. The panel concluded that the agency’s reasons for doubting that the individuals were police officers was not supported by substantial evidence.

Because the record compelled the conclusion that the two individuals involved in De Leon’s attack were police officers, the panel concluded that the agency’s determination that De Leon did not establish government acquiescence could not stand. The panel explained that government acquiescence is not required when instances of past torture are directly inflicted by a public official, even if that official has gone “rogue” by acting outside his or her authority. Moreover, the panel noted that the two individuals not only participated in De Leon’s beating, but also acquiesced during that incident in the actions of the other assailants by failing to use their official authority to prevent De Leon from being repeatedly stabbed.

In light of the ambiguity in the BIA’s conclusion with respect to the question whether De Leon was subjected to past torture and the absence of any explanation for a conclusion on that question, if made, the panel remanded that issue for further consideration. The panel also concluded that the agency’s analysis of the probability that De Leon would be subjected to future torture disregarded several critical factors, including myriad contextual considerations pertinent to the likelihood of future torture, and so must be redone.

Dissenting, Judge Collins wrote that the majority’s decision resurrected many of the same flawed legal standards that the court previously applied in immigration cases, and that were expressly rejected by the Supreme Court in its unanimous decision in Garland v. Ming Dai , 141 S. Ct. 1669 (2021). First, by taking as credible every detail in De Leon’s testimony, even though the IJ found only that De Leon testified in a “generally” or “essentially” credible manner, the majority disregarded the principle that the agency is free to credit part of a witness’ testimony without necessarily accepting it all. Second, even if the majority were right in thinking that every detail of De Leon’s testimony should be deemed credible, the majority further erred by effectively reviving the “deemed true” rule. And third, the majority disregarded the Supreme Court’s clear instruction that the court may not overturn a factual finding if the record contains contrary evidence of a kind and quality that a reasonable factfinder could find sufficient.

COUNSEL

Karla L. Kraus (argued), Kraus Law Corporation, San Diego, California, for Petitioner. John B. Holt (argued), Juria J. Jones, and Scott M. Marconda, Trial Attorneys; Claire L. Workman, Senior Litigation Counsel; Leslie McKay, Assistant Director; Jeffrey B. Clark, Acting Attorney General; Civil Division, Office of Immigration Litigation, Washington, D.C.; for Respondent.

OPINION

BERZON, Circuit Judge: Risvin Valdemar De Leon Lopez (“De Leon”), a native and citizen of Guatemala, petitions for review of a Board of Immigration Appeals (“BIA”) decision dismissing his appeal of an Immigration Judge’s (“IJ”) order denying his application for relief under the Convention Against Torture. We conclude: (1) the record in this case compels the conclusion that two of De Leon’s attackers were police officers during a July 2011 incident; (2) De Leon showed acquiescence on the part of the Guatemalan government with respect to that incident because government officials namely, the two police officers directly participated in the incident; and (3) the record indicates that the IJ and BIA’s conclusion that De Leon is not likely to be subjected to torture with government acquiescence if returned to Guatemala disregards several important circumstances pertinent to evaluating the likelihood of future torture. In light of these errors, we grant the petition and remand for the agency to reconsider De Leon’s application for relief.

I.

De Leon entered the United States without inspection in 2003 and stayed until the middle of 2007. He returned to Guatemala then because his oldest daughter was sick and stayed until the end of the year. De Leon next entered the United States, again without inspection, in 2008. He stayed until he pleaded guilty to driving under the influence and was removed, in 2011. De Leon entered the United States without inspection once more on January 5, 2012. He testified that on his return to Guatemala in 2011, he experienced two episodes of violence at the hands of the Guatemalan police. D E L EON L OPEZ V . G ARLAND

A.

De Leon testified as to what happened to him in Guatemala as follows: 1. De Leon was attacked in July 2011, two days after he arrived back in Guatemala. Earlier that day, De Leon had left his home in Aldea Galvez, a small village that is part of a larger town, Flores Costa Cuca, to visit a park with two friends. The three of them returned on a bus that afternoon. As they stepped off the bus, De Leon and his friends were approached by Melvin Baten and Elder Ramos. Baten and Ramos knew De Leon had just returned from the United States and so asked him for money. When De Leon refused, he was attacked by Baten and Ramos; Israel Augustin Alvarado, Oljoel Pascual Gomez, [1] and Minor Rojas also participated in the attack. At the time of the attack, Alvarado and Gomez were dressed in the uniform of the National Civil Police (“PNC”), carried handguns consistent with those of PNC officers, and were known by the Aldea Galvez community to be police officers.

The assailants started by “throwing rocks at [De Leon] with slingshots.” De Leon was hit in the chest and legs. When he bent down from pain, the attackers came towards De Leon and “stepped on [his] right hand.” They beat him with their fists and kicked him. Baten then “stabbed [De Leon] with a knife on the hand” and Ramos stabbed De Leon “with a machete on [his] right arm” and his “left shoulder.” The assailants took De Leon’s money and told him “we are going to kill yo u.” Alvarado and Gomez never tried to stop the other men from stabbing De Leon and actively participated in the beating. De Leon tried to defend himself, but Gomez hit him on the head with either a gun or baton, making De Leon lose consciousness. De Leon was later told by a relative that he was unconscious for half an hour.

De Leon knew all five men before the incident and, according to the IJ, “identified [them] in great detail” in his testimony. The men lived in the same small town as De Leon; some had gone to school or worked with him. With respect to Alvarado and Gomez, De Leon had known each of them for around twenty years before the incident. De Leon testified that he knew Alvarado and Gomez worked for the police in the Flores Costa Cuca station because of “conversations [De Leon] had with people” in his small town, w here “everybody knows what’s going on.”

De Leon’s friends, who had run away, returned to the scene of the incident along with De Leon’s aunt and uncle; the relatives thereafter witnessed part of the attack. De Leon’s aunt called the police. By the time the police arrived, the five assailants had left. The summoned police officers spoke with several witnesses at the scene of incident and later filed a report detailing the location of the incident and the names of De Leon’s attackers. The names were giv en to the police by persons at the scene, who “didn’t give [their own] names for fear of [retaliation].”

The police report said that the witnesses had identified “the individuals” who attacked De Leon as members of the 18th Street Gang. De Leon clarified in his testimony that three of the men (Ramos, Baten, and Rojas) were members of that gang. He also stated that he heard from the neighbors in his small town that Alvarado and Gomez were also “involved together with the gangs” in an extortion ring targeting local small businesses. De Leon testified that he D E L EON L OPEZ V . G ARLAND never saw the assailants again; he believed they may have gone to Mexico.

That day, De Leon was taken to a hospital in Coatepeque. He received stitches on his right bicep, seven stitches on his left shoulder, eight stitches on his hand, and sixteen stitches on his head. De Leon also had “two tendons cut from [his] three fingers” that “were stitched internally,” and cuts on his neck. The doctors prescribed medication for pain and to prevent inflammation, infection, and tetanus.

De Leon went to a clinic close by Flores Costa Cuca three times afterward for wound treatment. He could not work for the first three months after the attack, and it took approximately one year for him to recover physically from his injuries. De Leon also “consulted with a psychologist,” and “complained of having suffered psychological damage, such as difficulty sleeping, memory loss, [and] panic attacks.” His psychological trauma persisted at the time of the hearing.

Aft er two months passed, De Leon realized that “nothing was done” about the attack: the police “didn’t go after the people that had beaten [him]” and his assailants “were never caught.” Convinced that the police would not follow up on the report from July without being prompted, De Leon in September 2011 again reported the attack to the PNC of Flores Costa Cuca, the same office where Alvarado and Gomez worked as police officers. After De Leon told an officer at the Flores Costa Cuca station that two of their own officers had attacked him, the officer stated “[t]hat they couldn’t deal with [his] case there,” but told De Leon he had “a very strong case, that it was attempted murder, [and] that [he] should go to . . . [the] public ministry.” De Leon understood the officer’s reference to “very strong case” as indicating that the Flores Costa Cuca police “would be incapable of dealing with [his] kind of case[]” as it was “too difficult” and required “a deeper investigation.” De Leon emphasized his belief that th e police were “corrupt,” that they “let themselves be bribed very easily,” and that he doubted the coworkers of Alvarado and Gomez would be willing to take action against them.

As advised by the local police office, De Leon went to the Public Ministry of Coatepeque, a prosecutorial and investigative government agency semi-independent from the Guatemalan executive branch, to report the incident. His statement was taken and an investigation opened. The Public Ministry informed De Leon that Ramos, Baten, and Rojas were already under investigation because other complaints had been filed against them. The record also suggests that the Public Ministry indicated to De Leon that Alvarado and Gomez left the police force shortly after the attack. Specifically, the following exchange occurred between De Leon and the IJ:

IJ to De Leon: So after you were beaten, the two men involved who wore the police uniforms, stopped working for the police, as far as you know. De Leon to IJ: Yes. IJ to De Leon: And how do you know that? How did you learn that? De Leon to IJ: The public ministry tried to investigate them. IJ to De Leon: And they had already left their police job. De Leon to IJ: That’s right. D E L EON L OPEZ V . G ARLAND

De Leon interacted with officials from the Public Ministry about three times.

In addition, De Leon filed a complaint with the Office of the Attorney General. He believed the Office issued warrants against the five assailants. De Leon met with officials from the Attorney General’s Office four times. To De Leon’s knowledge, as of J anuary 2018, his assailants had not been tracked down and no one had been taken into custody for his assault.

2. De Leon had a second incident with Guatemalan police. In October 2011, De Leon went with his mother to Mexico, about an hour and a half from his home, to purchase goods for resale in her small store in Guatemala. Their return bus was stopped by police officers. The officers told De Leon and his mother to get out of the bus because “they wanted some money, [or] if not they were going to take the merchandise.” When De Leon refused, the officers handcuffed him and took him to the central police station.

At the station, De Leon was brought to a room where three officers took turns beating him with their hands and batons and kicking him with their feet. He was detained for around two to two and a half hours. Before leaving, De Leon told the officers “that [he] was going to the human rights [organization]” to report them. The police officers then told De Leon that “if [he] went to the human r ights organization that [he] was going to have problems with them.”

De Leon bought acetaminophen and was injected with an anti-inflammatory medication for his injuries. The next day, De Leon made a report to a government-affiliated human rights organization in Coatepeque. The organization took his statement and said they were going to do an investigation. But they never followed up with De Leon, and he never received “any information about them doing any type of investigation.”

3. Afraid to stay in Aldea Galvez after the police said “[he] was going to have problems with them” for making the human rights report, De Leon went to Guatemala City in December 2011. While he was in Guatemala City, two men carrying weapons went to his uncle’s house in Aldea Galvez “and asked [about De Leon’s] whereabouts.” Two armed men also went to the agricultural field where De Leon had worked and to the cemetery near his home, asking for him. De Leon feared that these men were the officers who had beaten him in October 2011 and that they wanted to carry out their earlier threats and murder him, as he had made a human rights report against them. De Leon stayed in Guatemala City for about one month and did not “have any problems” there.

De Leon left Guatemala in January 2012 because he was afraid of being “persecuted . . . or tortured or killed” by police officers. He testified that as he was on his way to the United States, his mother called and informed him that three armed police officers had come to her home “at 1: 00 in the morning” looking for him. The officers told De Leon’s mother they were searching for De Leon and “they were going to come back and look for [him] later.”

B.

In June 2015, about three and a half years after De Leon arrived back in the United States, he was detained by Immigration and Customs Enforcement (“ICE”) after a police call concerning an argument with his domestic partner. ICE arrested De Leon and issued a Notice of Intent/Decision to Reinstate Prior Order based on his June 2011 removal. De Leon expressed a fear of returning to D E L EON L OPEZ V . G ARLAND Guatemala and was interviewed by an asylum officer. The asylum officer found that De Leon had “established a reasonable fear . . . that he could be tortured if he’s returned” to Guatemala.

De Leon was then referred to an IJ and applied for relief under the Convention Against Torture. At the June 2018 merits hearing before the IJ, De Leon testified to the facts recited above about the two 2011 incidents. He averred that, to his knowledge, the five men from the July 2011 incident had not returned to Guatemala, but “it would be very easy for them to find out” if De Leon returned to their small town, because their family members live there. De Leon also testified to his fear that the five men would seek to “elimin ate him” upon his return, because “they would believe that [he] would continue with [his] case” against them. De Leon did not believe that the police would protect him from any attacks because, he stated, the police in Guatemala tend to be corrupt and generally neglect their duties. De Leon also noted that the police accept bribes, and he felt it would be “likely” or “probable” that the attackers’ families would bribe the police to keep them from protecting De Leon.

De Leon also voiced his fear that the police from the October 2011 incident would seek to “kill him” if he returned to Guatemala, “because they are aware that he had reported them to the human rights organization.” He noted that he would not feel safe even if he moved to another city in Guatemala such as the capital city, because when corrupt officers “really want you— when they want to go after someone they go to Guatemala [City] and they go look for you.”

During the merits hearing, the IJ initially would not accept the latest country conditions report for Guatemala on the ground that De Leon’s counsel should have filed it with the court earlier. After De Leon’s counsel pointed out that the “regulations say that the court must refer to” the country conditions report, the IJ accepted the country conditions report, moments before rendering her oral decision.

The country conditions report states that one of the “[p]rincipal human rights abuses” in Guatemala is “widespread institutional corruption, particularly in the police” force. It indicates that “abuse and mistreatment” of civilians at the hands of PNC members is common; that there are “credible reports of extrajudicial arrests [and] illegal detentions”; that the Guatemalan police frequently accept bribes or extort civilians; and that the government lacks effective mechanisms to investigate and prevent such wrongdoing. Finally, the report indicates that non-corrupt police officers commonly neglect to investigate or punish their corrupt peers beyond merely “transferr[ing]” them to other offi ces, which has led to widespread “impunity” for corrupt officials.

At the conclusion of the merits hearing, the IJ found that “[e]ssentially [De Leon] testified in a credible manner” but “ha[d] not established that it is more likely than not that he will be tortured with the consent or acquiescence of the authorities in Guatemala” should he be returned there. The IJ concluded that his “testimony was based in significant part on his speculation of who harmed him and why they harmed him and who might harm h im in the future.” “Speculation alone,” the IJ stated, “isn’t sufficient to establish respondent’s case.”

De Leon appealed to the BIA. The BIA agreed with the IJ that De Leon’s “assertion that several of the individuals who harmed him in Guatemala were police officers seems to be based on his own speculation.” The BIA dismissed De Leon’s appeal, concluding that: (1) the two attackers who D E L EON L OPEZ V . G ARLAND appeared to be police officers were not actually police officers; (2) even if the two attackers were police officers, “the record does not indicate that they tortured [De Leon] and that the government of Guatemala acquiesced to such torture”; and (3) “the record indicates that [De Leon] could . . . live in Guatemala City or another part of Guatemala without any harm.” The BIA did not mention the country conditions report. This petition for review followed.

C.

“Where, as here, the BIA agrees with the IJ’s reasoning, we review both decisions.” Garcia-Martinez v. Sessions , 886 F.3d 1291, 1293 (9th Cir. 2018). The BIA’ s interpretation of legal questions is reviewed de novo. Zheng v. Ashcroft , 332 F.3d 1186, 1193 (9th Cir. 2003). With regard to the factual findings underlying an IJ or BIA determination, the court reviews for substantial evidence, meaning that the determination must be supported by “reasonable, substantial, and probative evidence on the record.” Lopez v. Sessions , 901 F.3d 1071, 1074 (9th Cir. 2018). If any reasonable adjudicator would be compelled to conclude to the contrary of the IJ or BIA based on the evidence in the record, then the finding is not supported by substantial evidence. Garland v. Ming Dai , 141 S. Ct. 1669, 1677 (2021) (citing 8 U.S.C. § 1252(b)(4)(B)). But if the record contains evidence supporting the IJ or BIA’s conclusion and a reasonable factfinder could find that evidence sufficient, a reviewing court cannot overturn the agency’s factual determination. Id.

Several additional points are worth noting about the substantial evidence standard. First, although the standard is “highly deferential,” id. , “deference does not mean blindness,” Garcia v. Wilkinson , 988 F.3d 1136, 1142 (9th Cir. 2021) (quoting Parada v. Sessions , 902 F.3d 901, 909 (9th Cir. 2018)), and no deference is due to “inference[s] drawn from facts which are uncertain or speculative and which raise only a conjecture or a possibility,” Cal. State Water Res. Control Bd. v. FERC , 43 F.4th 920, 930 (9th Cir. 2022) (quoting Woods v. United States , 724 F.2d 1444, 1451 (9th Cir. 1984)). Second, the substantial evidence standard requires review of the record “as a whole,” not a “specific quantum of supporting evidence.” Garrison v. Colvin , 759 F.3d 995, 1009 (9th Cir. 2014) (quoting Lingenfelter v. Astrue , 504 F.3d 1028, 1035 (9th Cir. 2007)); see also Bringas-Rodriguez v. Sessions , 850 F.3d 1051, 1059 (9th Cir. 2017). Third, applying the substantial evidence standard after the Supreme Court’s decision in Ming Dai , the Ninth Circuit has set aside the BIA’s factual findings when the basis for the findings was insufficient or illogical. See, e.g. , Gonzalez-Castillo v. Garland , 47 F.4th 971, 979 80 (9th Cir. 2022); Ballinas-Lucero v. Garland , 44 F.4th 1169, 1180 (9th Cir. 2022); Barseghyan v. Garland , 39 F.4th 1138, 1145 46 (9th Cir. 2022); Plancarte Sauceda v. Garland , 23 F.4th 824, 835 (9th Cir. 2022); Munyuh v. Garland , 11 F.4th 750, 764 (9th Cir. 2021); Vasquez-Rodriguez v. Garland , 7 F.4th 888, 898 99 (9th Cir. 2021).

II.

The Convention Against Torture prohibits the United States from returning anyone to a c ountry where “it is more likely than not that he or she would be tortured.” 8 C.F.R. § 1208.16(c)(2). In evaluating a Torture Convention claim, “the IJ must consider all relevant evidence; no one factor is determinative.” Maldonado v. Lynch , 786 F.3d 1155, 1164 (9th Cir. 2015) (en banc). Relevant evidence includes, but is not limited to: (1) “evidence of past torture inflicted upon the applicant”; (2) “evidence that the applicant could relocate to a part of the country of removal where he or she is not l ikely to be tortured”; (3) “evidence of gross, flagrant D E L EON L OPEZ V . G ARLAND or mass violations of human rights within the country of removal, where applicable”; and (4) “other relevant information regarding conditions in the country of removal.” 8 C.F.R. § 1208.16(c)(3) (capitalization simplified).

Here, several of the IJ and BIA’s key factual findings and legal conclusions regarding whether De Leon was subject to past torture are not supported by the record, and the BIA’s determination regarding the likelihood of future torture disregards or mis-analyzes several pertinent considerations, including the Guatemala country conditions report and the threats of future harm made against De Leon.

A.

1. The IJ and BIA concluded that De Leon had not established that his attackers during the July 2011 incident were police officers. Instead, “these individuals wore police uniforms or uniforms that made them look like police uniforms but they were not . . . actually police employees”— i.e., they were imposters. Although the IJ found De Leon “credible” and did not doubt the sincerity of his belief that two of his attackers were police officers, the IJ and the BIA concluded that De Leon’s belief was “based on his own speculation” and was “not supported” by the police report. Those assessments are not supported by substantial evidence.

De Leon presented four significant pieces of evidence that two of his attackers, Alvarado and Gomez, were police officers. First, De Leon had known Alvarado and Gomez for about twenty years at the time of the incident. He had been told that the two men were police officers by people in his small town who were likely to know whether they were, in fact, police officers. Second, Alvarado and Gomez were wearing police uniforms at the time of the attack. Third, Alvarado and Gomez were armed with visible handguns that De Leon recognized as of the kind and in the holstered position typical of PNC officers. Fourth, the Public Ministry in Coatepeque indicated to De Leon that Alvarado and Gomez had left the police force shortly after the attack, thus confirming that they were members of the PNC during the July 2011 incident. Nothing in the IJ or BIA opinions suggests that De Leon was inaccurate or mistaken with regard to any of these factors.

The IJ and BIA nonetheless concluded that Alvarado and Gomez were not actually police officers even though when they attacked De Leon they were dressed and carried guns as if they were. The record as a whole cannot be reasonably understood to support this conjecture. See Garrison , 759 F.3d at 1009.

First, the IJ relied upon the police report about the July 2011 incident, which identified “the individuals” who attacked De Leon as associated with a street gang, for its finding. The report does not support the IJ’s speculation. The police report summarized what witnesses to the incident said; it did not specifically name Alvarado and Gomez as gang members; and it did not say that Alvarado and Gomez were not police officers. It is, of course, possible for law enforcement officers to be associated with a street gang. De Leon testified that, according to the neighbors in his small town, Alvarado and Gomez were “involved together with the gangs” in an extortion ring targeting small stores. The country conditions report indicates that such extortion is widespread in Guatemala. On the record as a whole, the absence in the police report of a statement that the two individuals dressed as police officers were police officers is not substantial evidence for the IJ’s finding. D E L EON L OPEZ V . G ARLAND

Second, the IJ characterized De Leon as “chang[ing] . . . his testimony” with respect to whether Alvarado was a police officer at the time of the attack. In its briefs to us, the Government, going one step further, states that De Leon “repeatedly testified that [Alva rado and Gomez] were former police officers.”

De Leon never used the phrase “former police officers.” He consistently explained in his 2018 testimony that both Alvarado and Gomez were police officers at the time of the July 2011 incident. For example, when directly asked if after you were beaten, the two men involved [in the July 2011 incident] who wore police uniforms, stopped working for the police,” De Leon responded, “Yes.” And when De Leon was specifically asked by the IJ when “one of th e men, Israel, used to work for the police,” he said “[i]n 2011 and before then.”

The IJ focused on De Leon’s occasional use of the past tense during his testimony, suggesting that De Leon was inconsistent about whether Alvarado and Gomez were police officers at the time they attacked him. What De Leon said at his hearing was that Alvarado “used to work with the police” and Gomez “also was working in the police.” De Leon was testifying about an incident that had occurred about seven years earlier, and Alvarado and Gomez had left the police force after the incident. So, at the time of De Leon’s testimony, it was accurate for him to say that the two men used to work with the police. The IJ’s grammatical dissection is both flatly incorrect, given the time frames, and contradicted by De Leon’s specific explanation as to what he meant by “used to work with the police.” Again, the IJ’s inference is simply not supported by the factual record.

Finally, the IJ stated that De Leon “offered no cogent explanation of why the police would take a report, list these individual’s names and then pretend they weren’t police if in fact they were police officers.” She further remarked that “[i]t’s clear if these two men had been police officers who were actively working for the forces the authorities would know that and they would be easy to identify and locate.” Additionally, the IJ pointed out that the attackers from the July 2011 incident fled the country, “indicating that they feared the police.” The IJ ultimately conc luded that Alvarado and Gomez must have “wor[n] police uniforms or uniforms that made them look like police,” but that “they were not in fact actually police employees.”

Regarding the omission of the attackers’ occupations, the police report did not mention the occupation of any of the five attackers. So the omission of Alvarado’s and Gomez’s occupation does not stand out as notable. Also, De Leon did offer an explanation as to why the authorities would decline to identify Alvarado and Gomez as police officers in the police report that officers in Guatemala commonly neglect to investigate or punish their peers for purported wrongdoing, preferring instead to look the other way. The country conditions report supports De Leon’s explanation.

Regarding the IJ’s suggestion that if Alvarado and Gomez were police officers they would be easy to track down, the record indicates that Alvarado and Gomez left the police force and fled the country shortly after the July 2011 altercation. That Alvarado and Gomez left the country after the attack supports an inference that they feared the police but does not support the inference that they were not police officers at the time of the attack. Corrupt police officers can fear that they will be caught and prosecuted.

In addition, the IJ and BIA cite no evidence that gang members in Guatemala dress up as police officers when D E L EON L OPEZ V . G ARLAND extorting civilians. The idea that they would have done so during the July 2011 incident is undermined by a fact that the IJ and BIA glossed over De Leon’s attackers from the second incident in October 2011, who were also wearing police uniforms, handcuffed him and brought him to the central police station before they took turns beating him with their hands, batons, and feet. That De Leon was beaten and detained in a police station in October 2011 provides strong evidence that his attackers from that incident were law enforcement officers, thereby confirming that police officers in De Leon’s area of Guatemala engage in beatings and extortion. [2] Again, the substantial evidence standard requires consideration of the record as a whole, see Garrison , 759 F.3d at 1009; here, any inference that a beating by actual police officers was improbable is belied by the evidence that such a beating had actually occurred, in a police station.

One final point regarding whether Alvarado and Gomez were police officers: At oral argument, there was a suggestion that, even if Alvarado was a police officer, Gomez may not have been, because De Leon indicated in his 2018 testimony that, at one time, he and Gomez had worked together “harvesting coffee at a farm.” When De Leon was asked what Gomez “did for a living at the time [De Leon] returned to Guatemala in 2011,” De Leon stated that, like Alvarado, Gomez “also was working in the police.” De Leon explained he knew this not only because Gomez was wearing a PNC uniform on the day of the July 2011 incident, but also because Gomez “live[d] in the same small village,” where “everybody knows what’s going on.” That De Leon and Gomez worked together at a coffee farm some time in the past does not supply the missing evidence sufficient to find that Gomez was not an officer at the time of the July 2011 attack. [3]

We conclude that the record compels the conclusion that two of De Leon’s attackers were police officers during the July 2011 incident. There is not substantial evidence in the record really, no evidence that Alvarado and Gomez were faux police officers. Put differently, no reasonable factfinder could have concluded that Alvarado and Gomez were imposters, dressed up like police officers but not actually employed in the PNC police force, when they attacked De Leon.

2. The dissent, citing to parts of the administrative record not discussed at all in the IJ or BIA decisions or in the briefing before this Court, argues that “this is most definitely not a case in which the record compels acceptance of ” De Leon’s statement that he had known Gomez and Alvarado for twenty years and knew that they were police officers at the time of the attack. Dissent at 37 40. The evidence cited by the dissent to challenge De Leon’s credibility with respect to the identity of the police officers cannot be the basis to sustain the IJ’s decision, for three independent reasons.

First, the IJ’s credibility determination forecloses the dissent’s own conclusions as to whether De Leon was testifying to what he believed to be true. As Ming Dai instructs, the Immigration and Nationality Act distinguishes between “credibility” and “persuasiveness.” 141 S. Ct. at 1680 81. Although the IJ did not find the entirety of De D E L EON L OPEZ V . G ARLAND Leon’s testimony persuasive, she repeatedly concluded that De Leon’s testimony was credible in the sense that De Leon believed that two of these men were police officers.” The IJ noted that De Leon “identified [his attackers] in great detail,” “stated . . . [that] they were wearing police uniforms,” and “testified that he has known them for a long time.” [4]

“We may not supply a reasoned basis for the agency’s action that the agency itself has not given.” Motor Vehicle Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co. , 463 U.S. 29, 43 (1983). The dissent’s reliance on evidence in the record that the government or the IJ could have raised to challenge De Leon’s credibility but did not amounts to an impermissible judge-made negative credibility finding that directly conflicts with the agency’s own determination. See Dissent at 37 40 . But “the IJ is in the bes t position to assess the trustworthiness of the applicant’s testimony,” Bassene v. Holder , 737 F.3d 530, 536 (9th Cir. 2013), and its credibility determinations are accorded “special deference,” Khadka v. Holder , 618 F.3d 996, 1000 (9th Cir. 2010). The di ssent’s attempt to undermine De Leon’s credibility with portions of the record never alluded to by or before the agency is improper.

Second, under well-established Ninth Circuit precedent, De Leon cannot be denied relief due to any alleged inconsistencies in his testimony absent an opportunity to respond to them. “If the IJ relies upon purported inconsistencies to make an adverse credibility determination, the IJ must provide the noncitizen with an opportunity to explain each inconsistency . . . .” Barseghyan v. Garland , 39 F.4th 1138, 1143 (9th Cir. 2022); see also Rizk v. Holder , 629 F.3d 1083, 1088 (9th Cir. 2011), overruled in part on other grounds by Alam v. Garland , 11 F.4th 1133 (9th Cir. 2021) (en banc). De Leon was not provided with an opportunity to explain the alleged inconsistencies the dissent raises for the first time in this case. So, even if otherwise pertinent, the evidence cited by the dissent could not be used to deny relief to De Leon under binding precedent.

Third, at no point in the Government’s brief did it challenge De Leon’s credibility or raise the alleged inconsistencies offered by the dissent. “Generally, an appellee waives any argument it fails to raise in its answering brief.” United States v. Dreyer , 804 F.3d 1266, 1277 (9th Cir. 2015) (en banc). None of the exceptions to the waiver rule apply in this case; in fact, accepting the dissent’s sua sponte credibility attack without offering De Leon an opportunity to respond would create not prevent a “miscarriage of justice.” Id.

3. The IJ and BIA reasoned that even if the July 2011 beating of De Leon constituted torture, the record does not indicate that the government acquiesced in such torture. That determination cannot stand if, as we conclude, the record compels the conclusion that two of the attackers were police officers. An applicant for relief under the Convention Against Torture is “not required to show acquiescence” by the government when instances of past torture are directly inflicted by a public official, Avendano-Hernandez v. Lynch , 800 F.3d 1072, 1079 80 (9th Cir. 2015), even if that official has gone “rogue” by acting outside his or her authority, Barajas-Romero v. Lynch , 846 F.3d 351, 362 (9th Cir. 2017). D E L EON L OPEZ V . G ARLAND

Alvarado and Gomez directly inflicted severe pain and suffering on De Leon by participating in the July 2011 beating. They also acquiesced during that incident in the actions of the other assailants by failing to use their official authority to prevent De Leon from being repeatedly stabbed with a knife and a machete. See Ornelas-Chavez v. Gonzales , 458 F.3d 1052, 1059 (9th Cir. 2006) (citing 8 C.F.R. § 208.18(a)(7)). As the record compels the conclusion that Alvarado and Gomez were police officers at the time of the incident, De Leon need not show any other acquiescence on the part of the Guatemalan government. Id.

4. The Torture Convention defines torture as “any act by which severe pain or suffering, whether physical or mental, is intentionally inflicted on a person for such purposes as . . . punishing him . . . for an act he . . . has committed . . . when such pain or suffering is inflicted by, or at the instigation of, or with the consent or acquiescence of, a public official acting in an official capacity or other person acting in an official ca pacity.” 8 C.F.R. § 1208.18(a)(1). The regulation does not require that past torture or “the prospective risk of torture be on account of certain protected grounds.” Kamalthas v. INS , 251 F.3d 1279, 1280 (9th Cir. 2001). “Acts constituting torture are varied, and include beatings and killings.” Bromfield v. Mukasey , 543 F.3d 1071, 1079 (9th Cir. 2008) (citing Al-Saher v. INS , 268 F.3d 1143, 1147 (9th Cir. 2001)).

Here, Alvarado and Gomez participated in beating De Leon in a manner that the IJ charact erized as “vicious” and a PNC officer framed as “attempted murder.” During the beating, the five attackers told De Leon that they were going to kill him. At no point did Alvarado or Gomez try to stop the other men from repeatedly stabbing De Leon with a knife and a machete, and Gomez himself knocked De Leon unconscious by hitting him in the head with either a gun or baton. De Leon needed sixteen stitches on his head as a result of Gomez’s strike; he needed over thirty stitches in total; he was out of work for three months; and he did not physically recover from his various injuries for over one year. De Leon also suffered ongoing psychological damage. Precedent indicates that such a beating may be sufficiently severe to qualify as torture. See, e.g. , Xochihua-Jaimes v. Barr , 962 F.3d 1175, 1188 (9th Cir. 2020); Bromfield , 543 F.3d at 1079; Muradin v. Gonzales , 494 F.3d 1208, 1211 (9th Cir. 2007); Al-Saher , 268 F.3d at 1146 48.

Neither the IJ nor the BIA directly addressed whether the July 2011 incident was severe enough to qualify as torture. The BIA stated that “even assuming that some of [De Leon’s] attackers were actual police officers, the record does not indicate [(a)] that they tortured the applicant and [(b)] that the government of Guatemala acquiesced to such torture.” In light of: (1) the absence of an express finding on severity in the IJ’s opinion; (2) the lack of any explanation in the BIA’s opinion as to why the July 2011 beating fell short of the definition of torture in terms of its severity; (3) the conjunctive structure of the BIA’s sentence; and (4) the BIA’s legally erroneous view, discussed above, that there was an inadequate showing of governmental acquiescence, it is most likely that the BIA’s holding was that, as the IJ held, the record did not establish acquiescence, and so did not establish (a) and (b), the conjunction necessary to demonstrate past torture.

At best, the BIA’s opinion is ambiguous as to whether it concluded that the July 2011 incident qualified as torture. If it did so conclude, the opinion contains no explanation for the holding.

In light of the ambiguity in the BIA’s conclusion with respect to the question whether De Leon was subjected to D E L EON L OPEZ V . G ARLAND past torture and the absence of any explanation for a conclusion on that question, if made, we remand that issue for further consideration. See Tadevosyan v. Holder , 743 F.3d 1250, 1257 58 (9th Cir. 2014); Arredondo v. Holder , 623 F.3d 1317, 1320 (9th Cir. 2010).

5. If, on remand, the agency determines that De Leon was subjected to past torture in July 2011, that conclusion would not fully satisfy De Leon’s burden to show that it is “more likely than not” that he would be tortured should he return to Guatemala. 8 C.F.R. § 1208.16(c)(2). Although “past torture is ordinarily the principal factor on which we rely when an applicant who has previously been tortured seeks relief under the Convention,” Nuru v. Gonzales , 404 F.3d 1207, 1218 (9th Cir. 2005), “changed circumstances” can lessen the importance of that factor. Avendano-Hernandez , 800 F.3d at 1080.

Here, circumstances have somewhat changed. Alvarado and Gomez are no longer members of the PNC and may no longer be in Guatemala. De Leon must therefore establish either that he would more likely than not be tortured in the future by Guatemalan officials or that the Guatemalan government would more likely than not acquiesce in future torture perpetrated against him by individuals who are not public officials.

The IJ and BIA concluded that De Leon did not make this showing. B ut “where there is any indication that [an IJ or] the BIA did not consider all of the evidence before it . . . the decision cannot stand.” Cole v. Holder , 659 F.3d 762, 771 –72 (9th Cir. 2011). The agency’s analysis of the probability that De Leon will be subjected to future torture disregarded several critical factors and so must be redone.

First, the BIA concluded that the second incident in October 2011 fell short of the definition of past torture because the beating and De Leon’s injuries were insufficiently severe. 8 C.F.R. § 1208.18(a)(1). De Leon does not contest the validity of that determination. But the October 2011 incident nonetheless qualifies as “relevant evidence” that the agency must consider in evaluating De Leon’s Torture Convention claim. Maldonado , 786 F.3d at 1164.

The record indicates that the police officers who beat De Leon in October 2011 told him that “if [he] went to the human rights organization” then he “was going to have problems with them.” Later, after De Leon made a human rights report, armed officers went to the homes of his relatives and to places he frequented looking for him, and the officers indicated that they would continue to “come back and look for [him] later.” [5] The October incident, threat, and follow-up make future torture by PNC officers more probable and should not have been ignored by the agency.

Second, “[t]he failure of the IJ and BIA to consider evidence of country conditions” in denying relief under the Torture Convention “constitutes reversible error.” Aguilar- Ramos v. Holder , 594 F.3d 701, 705 (9th Cir. 2010). The BIA’s decision in t his case does not reference the country conditions report at all. It does generally adopt the IJ’s reasoning. D E L EON L OPEZ V . G ARLAND

The IJ did not properly take into consideration the country conditions evidence De Leon proffered. That evidence demonstrates that many police officers in Guatemala are corrupt and involved in extortion, and that the officers who are not corrupt are unlikely to intervene to protect victims from harm. The IJ referred to the country conditions report only briefly in her decision. She stated that although the report “establish[es] there are some problems in Guatemala,” De Leon had not set forth a claim under the Convention Against Torture because “the authorities went out of their way to help him.”

The “problems” superficially referenced by the I J but recounted in detail in the country conditions report are directly relevant here. They include “widespread institutional corruption, particularly in the police” force; this corruption was characterized as one of the “[p]rincipal human rights abuses” in Guatemala by the country conditions report. The country conditions report demonstrates that the Guatemalan government lacks effective mechanisms to investigate and prevent abuse and corruption within the PNC, and that non-corrupt PNC officers are reluctant to punish the wrongdoings of their corrupt peers. The dismissive manner in which the IJ referred to and treated the country conditions report including characterizing it as “29 pages of general information of which we are all familiar,” and rendering her oral decision only moments after accepting the report into the record suggests that the IJ did not read the report in any detail or genuinely consider the parts directly pertinent to De Leon’s likely situation in Guatemala upon removal.

This conclusion is bolstered by Parada v. Sessions , 902 F.3d 901 (9th Cir. 2018). There, the IJ summarized the relevant country conditions in his decision, but there was a “significant and material disconnect between the IJ’s quoted observations and his conclusions reg arding [the petitioner’s Torture Convention] claim.” Id. at 915. That disconnect “indicate[d] that the IJ did not properly consider all of the relevant evidence before him.” Id . Here, the IJ did not quote or summarize the country conditions report in her decision. The engagement with the country conditions report in this case which was virtually non-existent on the part of the IJ, and completely non-existent on the part of the BIA is inadequate.

Third, the BIA suggested that De Leon could “live in Guatemala City or another part of Guatemala without any harm.” To find that an applicant can safely relocate to another part of the country, an IJ or the BIA needs to cite affirmative evidence supporting that determination. Xochihua-Jaimes , 962 F.3d at 1186 87 (citations omitted). The BIA’s finding is not supported by such evidence.

The Government argues that two pieces of evidence support the BIA’s determination that De Leon could safely relocate: (1) that De Leon lived in Guatemala City for approximately one month without being attacked, and (2) that De Leon’s attackers from the July 2011 incident have not returned to Guatemala to his knowledge. The parties do not cite, and we have not found, any cases holding that a single month’s stay in another region of the proposed country of removal qualifies, by itself, as sufficient affirmative evidence that the applicant could safely relocate to that region. That gap is not surprising. A one-month sojourn in another area of the proposed country of removal does not constitute substantial evidence that an applicant will not be found by his adversaries if he remains there permanently.

Additionally, both during the month De Leon lived in Guatemala City and later, armed police officers were D E L EON L OPEZ V . G ARLAND actively searching for him at his relatives’ homes and at locations he had previously frequented. The officers told De Leon’s relatives that they would continue to search for him. De Leon’s family membe rs did not disclose his location in Guatemala City to the armed officers. But police officers seeking to harm De Leon could uncover his location through other means, such as by asking other individuals living in De Leon’s small town or by contacting offic ials in other parts of the country.

Finally, although the five attackers from the July 2011 incident have not to De Leon’s knowledge returned to Guatemala from Mexico (the border of which is close to De Leon’s home town), he testified that news of his retu rn would travel through his small town quickly, and it would be “very easy” for potential attackers, including the aggrieved police officers involved in the October 2011 incident, to learn he had returned and track him down. The BIA’s assessment of whether De Leon could safely relocate should have taken into account these factors.

Fourth, Torture Convention “claims must be considered in terms of the aggregate risk of torture from all sources,” including from sources that did not have a hand in past torture inflicted upon the applicant. Xochihua-Jaimes , 962 F.3d at 1187 (quoting Quijada-Aguilar v. Lynch , 799 F.3d 1303, 1308 (9th Cir. 2015)). The BIA must assess whether, after “aggregating the risks posed” by each potential source of torture, the applican t has demonstrated “a probability greater than 50 percent that he will be tortured.” Velasquez-Samayoa v. Garland , No. 21-70093, 2022 WL 4453004, at *6 (9th Cir. Sept. 23, 2022). Here, the IJ and BIA exclusively considered the risk of harm to De Leon stemming from the particular individuals involved in the July 2011 incident. That approach ignores the risk that De Leon would be tortured in the future by other police officers, including those connected to the October 2011 incident or those who have similar motivations to the officers involved in the July and October incidents. Cf. Kaur v. Wilkinson , 986 F.3d 1216, 1230 31 (9th Cir. 2021).

The record indicates that a common theme underlies De Leon’s two violent encounters with the Guatemalan police. In both instances, De Leon refused to submit to corrupt police practices involving the theft of his money or goods. When asked why he did not simply submit to the corrupt practices to avoid being harmed, De Leon explained “it didn’t seem fair to [him]” to submit to “the kind of people that only want to harm others.” He also explained that he fundamentally “opposed” the corrupt practices “because that is no way to —they don’t have any right to take the merchandise from us or even ask for money.” The country conditions report indicates that Guatemalan police frequently extort citizens of the country. That evidence is pertinent to evaluating whether, if De Leon returned to Guatemala, his defiance against corrupt police practices could lead him to face new incidents of violence by or with the acquiescence of police officers unconnected to the July 2011 incident.

Additionally, the record demonstrates that De Leon was twice attacked by police officers on his way back to his small town. Both attacks were premised in part on the assumption that De Leon would be carrying money or other objects of wealth upon his return. The possibility that De Leon may be attacked in a similar fashion in the future if he were again removed to Guatemala from the United States should have been considered by the BIA in assessing the probability of future torture.

In sum, the IJ and BIA disregarded myriad contextual considerations pertinent to the likelihood of future torture. D E L EON L OPEZ V . G ARLAND To summarize those considerations: De Leon has already experienced two assaults at the hands of the police, not one; police officers warned him not to report them to a human rights organization and, after he made a report, searched for him at his family members’ homes and at places he frequented, suggesting that they intended to harm him again; and, given his principled resistance to police corruption, De Leon may anger additional corrupt officers in the future. Combined with the country conditions report’s material about police corruption, the record indicates that the violence De Leon experienced may not have been “isolated” or one-off. The BIA should have taken the broader picture into consideration but did not.

* * * When an IJ or BIA decision “cannot be sustained upon its reasoning,” the proper course of action on review is to remand for the IJ or BIA to reconsider its decision. Andia v. Ashcroft , 359 F.3d 1181, 1184 (9th Cir. 2004); see also Cole , 659 F.3d at 774. Here, the record compels the conclusion that two of De Leon’s attackers from the July 2011 incident were police officers, so he need not otherwise show acquiescence on the part of the Guatemalan government with respect to that incident. The determination whether De Leon was subjected to past torture therefore must be reconsidered. Separately, the IJ and BIA did not properly consider information pertinent to determining the probability of future torture when denying De Leon’s request for relief under the Torture Convention. We therefore grant the petition for review and remand the matter to the BIA for the agency to reconsider De Leon’s application for relief under the Torture Convention.

The petition for review is GRANTED . We REMAND for further proceedings consistent with this opinion. COLLINS, Circuit Judge, dissenting:

In rejecting Petitioner Risvin Valdemar De Leon Lopez’s claim for relief under the Convention Against Torture, the Immigration Judge (“IJ”) relied critically on an explicit factual finding that none of the five men who robbed and attacked De Leon in Guatemala in July 2011 were police o fficers. The Board of Immigration Appeals (“BIA”) upheld that finding and the resulting denial of relief, but the majority nonetheless reverses that factual finding and remands the case back to the agency. The majority’s opinion is a textbook example of how judicial review of immigration decisions ought not to be conducted. Indeed, the majority’s decision resurrects many of the same flawed legal standards that we previously applied in immigration cases and that were expressly rejected by the Supreme Court in its unanimous decision in Garland v. Ming Dai , 141 S. Ct. 1669 (2021). I respectfully dissent.

I

In addressing De Leon’s claim that, in July 2011, he was attacked by five men, including two police officers, the IJ reviewed the conflicting evidence and made an express factual finding that these two “individuals wore police uniforms or uniforms that made them look like police uniforms but they were not in fact actually police employees or any lawful authority.” Exercising review for clear error, the BIA expressly upheld this finding that the assailants were not “actual police officers.” As a result, under the INA, this finding of fact is “conclusive unless any reasonable adjudicator would be compelled to conclude to the contrary.” 8 U.S.C. § 1252(b)(4)(B). The majority nevertheless reverses this factual finding, holding that “the D E L EON L OPEZ V . G ARLAND record compels the conclusion that two of De Leon’s attackers from the July 2011 incident were police officers.” See Opin. at 28. In doing so, the majority seriously errs.

A

Before turning to the specifics of how the majority improperly reweighs the particular evidence in this case, I think it is important to note, at the outset, three respects in which the majority flouts the applicable principles of law that the Supreme C ourt set forth when it reversed this court’s decision in Ming Dai .

First, the Ming Dai Court told us that we were wrong to treat credibility determinations as an all-or-nothing matter, because an IJ, “like any reasonable factfinder, is free to credit part of a witness’ testimony without necessarily accepting it all.” 141 S. Ct. at 1677 (simplified). The majority disregards that principle by taking as credible every detail in De Leon’s testimony, even though the IJ found only that De Leon “testified in a generally credible manner” and an “ [e]ssentially ” credible manner (emphasis added). As the IJ made clear, she declined to accept all of De Leon’s testimony, because some portions were based on “speculation,” and other parts were “contradict[ed]” by his own documentary evidence and even his earlier testimony.

The majority contends that these comments by the IJ addressed only the “persuasiveness” of De Leon’s testimony and not its “credibility,” and on that basis the majority proceeds to treat De Leon’s testimony as having been found to be credible in all relevant respects. See Opin. at 17 18. The majority’s premise is incorrect. In particular, given that the IJ’s decision relies heavily on contradictions in and between De Leon’s testimony and past statements, that decision clearly reflects doubts as to De Leon’s credibility in claiming that his attackers were police officers. See infra note 1. By instead deeming all of De Leon’s testimony to be credible, the majority effectively revives the very “presumption of credibility” that Ming Dai told us does not apply on a petition for review in federal court. 141 S. Ct. at 1678. As the Court explained, we may not presume that any portion of an alien’s testimony that was not specifically rejected by the IJ was credited. On the contrary, even when the record is unclear as to whether testimony was explicitly disbelieved, the “only question for judges reviewing the BIA’s factual determinations is whether any reasonable adjudicator could have found as the agency did.” Id . By crediting De Leon’s testimony who lesale, and then deploying it to reject the agency’s findings, the majority “flips” this deferential “standard on its head” and thereby “gives conclusive weight to any piece of testimony that cuts against the agency’s finding.” Id .

Second, even if the majority were right in thinking that every detail of De Leon’s testimony should be deemed credible , the majority further errs by effectively reviving our further “deemed true ” rule, which Ming Dai also squarely rejected. See 141 S. Ct. at 1676 77. As I will show in discussing the majority’s analysis of the specific evidence in this case, the majority wrongly takes every detail of De Leon’s hearing testimony as the gospel truth, and it then aggressively discounts contrary evidence in the record by drawing d ebatable inferences in De Leon’s favor and contrary to the IJ’s factual findings. But as the Supreme Court admonished us, “even if the BIA [or the IJ] treats an alien’s evidence as credible, the agency need not find his evidence persuasive or sufficient to meet the burden of proof.” Id . at 1680. Here, the IJ clearly found that, even assuming that De Leon was credible in contending that two of his attackers wore uniforms, that testimony was not sufficiently persuasive to establish that they were in fact D E L EON L OPEZ V . G ARLAND police officers. Under Ming Dai , it was entirely proper for the IJ to conclude that “testimony on [that] key fact was outweighed by other evidence and thus unpersuasive or insufficient to prove” the alien’s case. Id . at 1681.

Third, the majority disregards Ming Dai ’s clear instruction that we may not “overturn” a factual finding if “the record contains contrary evidence of a kind and quality that a reasonable factfinder could find sufficient.” 141 S. Ct. at 1677 (simplified). As I will explain, the record clearly contains such evidence here. But the majority instead cherry-picks from the record whatever evidence it thinks supports its favored conclusion, while ignoring the very substantial evidence in the record that is contrary to that conclusion.

In defending its disregard of such contrary evidence, the majority asserts that, in applying the substantial evidence standard, we may consider additional items of record evidence that were not specifically mentioned by the IJ or the BIA only if those items undermine the agency’s conclusion. Compare Opin. at 13 17 (chastising the agency on the grounds that it overlooked items of evidence in the “record as a whole” that the majority thinks support a contrary view) with Opin. at 17 20 (arguing that we must c lose our eyes to record evidence supporting the agency’s conclusion if that evidence was not specifically cited by the agency). According to the majority, this one-way ratchet follows from the principle, derived from SEC v. Chenery Corp. , 332 U.S. 194, 196 (1947), that judicial review of an agency decision is limited to “only the grounds relied upon by that agency.” Andia v. Ashcroft , 359 F.3d 1181, 1184 (9th Cir. 2004). The majority’s reasoning is flawed. Here, the agency’s grounds for decision were clearly identified namely, that De Leon was not attacked by police officers and that his claims to the contrary were based on speculation and on testimony that was internally inconsistent and contradicted by other evidence in the record. Under the express terms of the INA, we may not set aside that finding and replace it with a contrary factual finding as the majority has done here —“unless any reasonable adjudicator would be compelled to conclude to the contrary,” 8 U.S.C. § 1252(b)(4)(B). Ming Dai states that, in applying this “highly deferential standard” to an agency’s explicit rejection of an alien’s factual contention, our task is simply to determine whether “the record contains contrary evidence of a kind and quality that a reasonable factfinder could find sufficient,” 141 S. Ct. at 1677 (simplified). Moreover, Ming Dai specifically rejects any “magic words” approach to reviewing the sufficiency of the evidence underlying the agency’s factual determinations. See id . at 1679. Accordingly, the majority ’s insistence that, in applying the substantial evidence standard to the explicitly stated grounds of the agency’s decision, we may consider only the particular snippets of record evidence specifically and expressly cited by the agency is a distortion of the Chenery rule, contravenes Ming Dai , and ignores the applicable statutory language.

B

The flaws in the majority’s legal standards are confirmed by the numerous mistakes the majority makes in assessing whether substantial evidence in the record supports the specific factual finding that is at issue in this case namely, whether two of De Leon’s attackers in the July 2011 incident were police officers. Even while crediting De Leon’s assertion that two of his attackers wore uniforms, the IJ gave four reasons for nonetheless concluding that the men were not police officers. [1] Collectively, these reasons provide D E L EON L OPEZ V . G ARLAND substantial evidence to support the IJ’s factual finding, and the majority errs in concluding otherwise.

First, the IJ concluded that, in discussing his knowledge about the occupations of the two men, De Leon at one point changed his testimony, first stating that one of his attackers used to work for the police” but then claiming that he “was working for the police at the time ” of the attac k (emphasis added). The majority attempts to explain away the inconsistency, arguing that De Leon’s use of the past tense in the first phrase was attributable to the fact that he “was testifying about an incident that had occurred about seven years earlie r.” See Opin. at 14. If I were the trier of fact, I might agree with the majority’s attempt to reconcile these seemingly conflicting aspects of De Leon’s testimony. But the IJ clearly read both statements as referring to the attackers’ status at the tim e of the attack and as therefore internally inconsistent. The majority’s adoption of a debatable contrary inference is flatly contrary to the “‘highly men were police officers.” See Opin. at 18 (quoting the IJ’s decision). This erroneous contention rests on an improperly truncated quotation from the IJ’s decision. In the cited sentence, the IJ observed that, According to his testimony respondent [De Leon] believed that two of these men were police officers and that the other three men were neighbors and people that respondent knew” (emphasis added). Far from endorsing the credibility of De Leon’s belief, the quoted sentence merely recounts that De Leon claimed such a belief. Moreover, the majority’s theory that the IJ supposedly found that De Leon actually believed the men to be officers is difficult to square with the fact that the IJ rejected De Leon’s testimony that the men were officers as being contrary to his own additional statements , both in the other portions of his testimony and in his earlier statement to investigators. A finding that, with respect to a particular point, a witness has given internally inconsistent testimony that also contradic ts the witness’s prior statements is not reasonably construed as an endorsement of the credibility of the testimony as to that particular point. deferential standard” of review. Ming Dai , 141 S. Ct. at 1677; see also Nasrallah v. Barr , 140 S. Ct. 1683, 1692 (2020). How to weigh potentially conflicting items of testimony is a discretionary judgment to be made by the IJ, subject to the BIA’s clear error review. We lack the authority to reweigh the evidence ourselves and to draw competing inferences. Indeed, even the BIA which operates under a less strict standard of clear-error review is forbidden to do what the majority has done here, which is to “rely on its own interpretation of the facts.” Guerra v. Barr , 974 F.3d 909, 912 (9th Cir. 2020).

Se cond, the IJ also reasoned that De Leon’s claim that the men were police officers was contradicted by the fact that, as soon as De Leon reported the attack, the men fled the country, “indicating that they feared the police” (emphasis added). The majority disputes the IJ’s inference, arguing that the men’s flight does not necessarily mean that they were not police officers at the time of the attack. As the majority contends, “[c]orrupt police officers can fear that they will be caught and prosecuted.” See Opin. at 15. Once again, the majority is choosing between debatable inferences, thereby flagrantly disregarding the standard of review. See Guerra , 974 F.3d at 913 (noting that, “[w]hen there are two permissible views of the evidence, the IJ’s choice be tween them” cannot be side aside, even by the

BIA).

Third, the IJ noted that the written copy of the police report concerning the incident did not mention that the men were police officers; instead they were identified as members of the 18th street gang. It was entirely reasonable for the IJ to conclude that this report undermined De Leon’s contention that two of the men were police officers. Yet again, the majority improperly substitutes its own tendentious reweighing of the evidence. According to the D E L EON L OPEZ V . G ARLAND majority, the report was only a summary of statements made by persons other than De Leon, and therefore no significance can be attached to its failure to mention that two of the men were police officers. See Opin. at 13. As an initial matter, the majority overlooks the fact that the report which is a summary of an interview with De Leon recounts De Leon’s statement that various persons, who were at the treatment center he went to after the attack, described the attackers as members of the “18 gang.” In ad dition, the majority overlooks De Leon’s subsequent statement to the public prosecutor’s office in which he stated that “these individuals that wounded me are members of the eighteen gang.” These various statements, whether by De Leon himself or by De Leo n relaying other persons’ claims, remain competing evidence that the IJ could consider in reaching a conclusion contrary to the majority’s.

The majority also holds that the agency should not have given weight to any perceived inconsistency on this score because, in any event, the men who attacked De Leon could have been both police officers and gang members. See Opin. at 13. De Leon made similar attempts to reconcile his testimony with the police report, but the IJ expressly rejected them, finding that De Leon “offered no cogent explanation of why the police would take a report, list these individual’s names and then pretend they weren’t police if in fact they were police officers.” In rejecting the IJ’s conclusion, the majority simply usurps the IJ’s a uthority, in direct contravention of the INA’s highly deferential standard of review.

Fourth, the IJ concluded that, because the authorities would know more about their own officers, their subsequent inability to locate the men was inconsistent with the conclusion that they were police officers. I confess that, if I were the trier of fact, I would not have drawn this inference, because it is presumably difficult to locate anyone who has fled the country, regardless of whether they are officers. But as I have already repeatedly observed, the decision as to which inferences to draw from the evidence belongs to the IJ and not to this court. The majority again disregards these constraints by instead adopting the competing inference that the men were not locat ed because “officers in Guatemala commonly neglect to investigate or punish their peers.” See Opin. at 15.

C

In addition to improperly reweighing and rejecting each of the specific inferences drawn by the IJ in support of her factual finding that the men were not officers, the majority also selectively rummages through the record in search of additional evidence to support its preferred conclusion. This cherry-picking overlooks the significant amount of additional contradictory evidence that seriously undermines De Leon’s claims and that show that this is most definitely not a case in which the record compels acceptance of his claims that two of the attackers were police officers. [2] D E L EON L OPEZ V . G ARLAND

For example, the majority chastises the IJ for overlooking De Leon’s tes timony that he had known the putative officers “for about twenty years” and that it was common knowledge in the village that they were police officers. See Opin. at 12. There is no indication, however, that the IJ credited this aspect of De Leon’s testim ony, which is not mentioned in the evaluative section of the IJ’s ruling. [3] Nor was the IJ compelled to do so, and nor are we. Indeed, the record contains ample “contrary evidence of a kind and quality that a reasonable factfinder could find sufficient” to justify declining to credit or mention this aspect of De Leon’s testimony. Ming Dai , 141 S. Ct. at 1677 (simplified). In particular, De Leon’s story about having known these two men to be police officers is very hard to square with the earlier contradictory statements he made during his removal proceedings. [4] testimony that the men were police officers, and we may not set that conclusion aside “so long as the record contains contrary evidence of a kind and quality that a reasonable factfinder could find sufficient.” Ming Dai , 141 S. Ct. at 1677 (emphasis added) (citations and internal quotation marks omitted). Moreover, as noted earlier, the majority’s distorted view of Chenery leads it to wrongly adopt an asymmetrical rule that uncited evidence may be used to attack the agency’s factual findings, but not to support them. See supra at 32 33. [3] The majority’s erroneous contrary contention is once again based on its patent misreading of the IJ’s decision. As before, see supra note 1, the majority misreads the IJ’s mere summary of what De Leon claimed in his testimony as reflecting an endorsement of the credibility of every claim mentioned. In each of the relevant sentences describing De Leon’s testimony about how long he knew the alleged officers, the IJ used phrasing such as “[a]ccording to his testimony”; “[r]espondent testified”; “[a]s respondent described him”; and “[r]espondent said”. [4] The majority claims it is unfair to consider these additional aspects of the record because, under our precedent, the agency supposedly could not have expressly relied on these particular contradictions without first specifically raising each one of them with De Leon and allowing him to

At the January 29, 2018 hearing before the IJ, De Leon named two men, Israel Augustin Alvarado and Oljoel Pascual Gomez, as the uniformed police officers who participated in the July 2011 attack. The majority simply assumes this testimony to be true, even though the IJ made no finding as to the identities of the two men. But in his declaration in support of his December 2015 application for relief, De Leon claimed that the two police officers who attacked him were “Joel Lopez” and “Marcelino, whom I don’t know his last name.” He further stated that the other attackers i.e. , the ones who were not police officers were “Israel Agustin Alvarado and Melvin Baten.” Similarly, in his June 24, 2015 reasonable fear interview, De Leon named the two police officers as “Joel Lopez” and a man named “Marcelino,” whose last name he did not know. In that interview, he was also asked “[w]hat were the names of the individuals who were not police officers,” and he replied, “Melvin Baten, Israel Augustine [ sic ] Alvarado.” These prior statements that Alvarado was not one of the officers and that De Leon did not know the last name of the other try to explain them. See Opin. at 18 19. Even assuming arguendo that our continued adherence to that “judge - made procedural requirement[]” is consistent with Ming Dai , but see 141 S. Ct. at 1677 (reminding us that judge- made rules “that Congress has not prescribed and the Constitution does not compel” have “no proper place in a reviewing court’s analysis”), it is inapposite here. I have raised these aspects of the record o nly in response to the majority’s selective rummaging of the record for additional evidence, not relied upon by the agency, that undermines the agency’s decision. The majority cannot have it both ways: we cannot simultaneously fault an agency for failing to rely on a particular aspect of the record without acknowledging the additional baggage with which that aspect is freighted. And even if the majority’s procedural rule were applicable here, it would at most warrant a reconsideration of the factual determination by the agency and not, as the majority would have it, a judge- made factual finding that is directly contrary to the agency’s. See , e.g. , Soto-Olarte v. Holder , 555 F.3d 1089, 1096 (9th Cr. 2009) (remanding on an open record). D E L EON L OPEZ V . G ARLAND officer thus directly contradict De Leon’s hearing testimony that Alvarado was one of the officers and that De Leon had known both men for 20 years.

Moreover, De Leon’s earlier statements are also inconsistent as to the number and identities of the non- police-officer attackers. In his July 18, 2011 statement to the Guatemalan prosecutor’s office, De Leon made no mention of any of the attackers being police officers, and he identified his attackers as “Joel Pascual Lopez Gomez,” “Elder R amos, Agustin Alvarado, and Melvin Baten.” In his declaration and credible fear interview, he also named four men, but he substituted the mysterious “Marcelino” for “Elder Ramos.” At the hearing, however, he named five men: Alvarado, Gomez, and Baten remained on the list, but Marcelino was dropped, Elder Ramos was added back, and a new fifth person was added Minor Rojas. As with Alvarado and Gomez, De Leon told the IJ he had known Rojas and Ramos for many years. The addition of Minor Rojas as a fifth at tacker was notable, because De Leon’s 2011 statement to Guatemalan prosecutors insisted that “Mynor Rojas” had absolutely nothing to do with the attack. [5]

With this mess of a record, we are certainly not compelled to accept De Leon’s assertion that he had known the men for 20 years and therefore knew they were police officers, and the IJ cannot be faulted for ignoring that claim.

II

Beyond its unwarranted conclusion that the attackers in the July 2011 incident included police officers, the majority does not otherwise identify any basis for setting aside the agency’s conclusion that the Guatemalan government had not acquiesced in any past alleged “torture” of De Leon. Nor does the majority contend that, in the absence of past torture, De Leon presented sufficient evidence to establish that future torture was likely if De Leon is removed to Guatemala. [6] Consequently, because the IJ’s factual finding that no police officers participated in the July 2011 attack should be upheld, De Leon’s claim for relief u nder the Torture Convention necessarily fails, and the petition should be denied.

* * * Because the majority revives and applies flawed legal standards that were unanimously rejected by the Supreme Court in Ming Dai , I respectfully dissent.

NOTES

[1] At times, the record refers to Gomez as Joel instead of Oljoel, and as Lopez or Lopez Gomez instead of Gomez. This opinion will refer to him as Gomez.

[2] The IJ found that the October 2011 incident did not rise to the level of past torture. The police officers’ extortion and beating of De Leon in October 2011 nevertheless provide support for his statement that other officers extorted and beat him in July 2011.

[3] In any event, it is sufficient for the purpose of determining whether De Leon suffered past torture at the hands of government officials that one such official, Alvarado, participated in the attack. See infra pp. 19 20.

[4] The dissent accuses us of “reviv[ing] the very ‘presumption of credibility’” rejected by Ming Dai . Dissent at 31. But our analysis rests on the IJ’s credibility finding regarding De Leon’s belief that two of his attackers were police officers, not on a presumption of credibility. The IJ’s conclusion that De Leon’s belief was based on “speculation” speaks to the persuasiveness of De Leon’s belief, not its credibility.

[5] The IJ and BIA did not dispute that the assailants from the October 2011 incident were police officers. The IJ did suggest that the armed officers who subsequently went to the homes of De Leon’s relatives and to places he previously frequented may have had a lawful purpose, such as to discuss the ongoing investigations against his attackers. That the armed officers approached the home of De Leon’s mother at 1:00 a.m. could belie that suggestion.

[1] The majority claims that the IJ made a specific finding that De Leon was credible when De Leon claimed that he “believed that two of these

[2] As noted earlier, the majority is wrong in contending that, in assessing whether the record evidence supports the IJ’s finding, we are limited by the Chenery rule to considering only those aspects of the record that were affirmatively cited by the IJ. In explaining how the Chenery rule applies within the context of the INA’s review provisions, Ming Dai explained that (1) in assessing whether particular testimony was rejected as not credible, it suffices “if the agency’s path may reasonably be discerned”; and (2) such a rejection must be upheld unless, reviewing the record as a whole, “a reasonable adjudicator wou ld have been compelled to reach a different conclusion.” 141 S. Ct. at 1679 (citation omitted); cf . Louisiana-Pac. Corp. v. NLRB , 52 F.3d 255, 258 59 (9th Cir. 1995) (distinguishing between the “basis” for an agency order, which is limited by the Chenery rule, and an agency “finding,” which is reviewed for “substantial evidence”). Here, the agency did not credit De Leon’s

[5] De Leon’s statemen t to prosecutors also stated that his companions, Noel Mendez and Aroldo Escobar, were attacked, which contradicted his hearing testimony that “[t]hey weren’t hit” and that “[t]hey just left the place.”

[6] The majority instead contends that, if on remand the agency concludes that De Leon’s past treatment amounted to torture, the agency must then redo its analysis of the likelihood of future torture. The agency’s prior discussion of that issue was flawed, according to the majority, because the agency failed to adequately consider several additional items of evidence that, together with a showing of past torture, may suffice to warrant relief here. This discussion also further reflects the majority’s flawed aggressive approach to reading the record in this case.