Third Circuit: CAT Deferral Requires Individualized Proof of Likely Torture—Arrest and “Ill-Treatment” Evidence Alone Is Insufficient

I. Introduction

In Javokhir Attoev v. Attorney General United States of America (3d Cir. July 6, 2026) (not precedential), the Third Circuit reviewed a final removal order after the Board of Immigration Appeals (“BIA”) reversed an Immigration Judge’s (“IJ”) grant of deferral of removal under the Convention Against Torture (“CAT”) and ordered removal to Uzbekistan.

The petitioner, Javokhir Attoev, entered the United States without authorization and was charged as removable under 8 U.S.C. § 1182(a)(6)(A)(i). He sought asylum, statutory withholding, and protection under CAT, claiming he had been targeted by Uzbek authorities based on his religious beliefs and would face persecution or torture if returned. The record also reflected an international warrant alleging terrorism involvement and domestic evidence of identity fraud, drugs, and a firearm. The IJ denied asylum and withholding (finding petitioner not credible), but granted CAT deferral based on a prediction of detention on return and a consequent risk of torture. The BIA affirmed the denials but reversed CAT deferral, concluding the IJ had conflated risk of arrest with risk of torture and that the evidence was insufficient to show the petitioner himself was more likely than not to be tortured.

On petition for review, Attoev argued (1) the BIA used the wrong standard of review by engaging in improper de novo factfinding, and (2) the BIA’s reversal lacked substantial evidence. The Third Circuit denied the petition.

II. Summary of the Opinion

The Third Circuit held that the BIA applied the proper standard of review: it did not disturb the IJ’s factual findings (including the finding that Attoev was likely to be detained upon return), but instead corrected a legal error—treating likely detention as equivalent to likely “torture” under CAT. The Court further held that substantial evidence supported the BIA’s conclusion that generalized country-conditions evidence and an ECHR decision referencing “ill-treatment” did not establish a clear probability that this petitioner would be tortured, particularly in light of the IJ’s adverse credibility finding and lack of individualized proof that Attoev would be targeted for torture.

III. Analysis

A. Precedents Cited

  • Toussaint v. Att'y Gen., 455 F.3d 409 (3d Cir. 2006)
    Cited for the standard that legal conclusions are reviewed de novo. The panel used this to frame its review of whether the BIA applied the correct standard to the IJ’s CAT deferral decision.
  • I.N.S. v. Elias-Zacarias, 502 U.S. 478 (1992)
    Cited for the substantial-evidence standard: agency factfinding is upheld if supported by “reasonable, substantial, and probative” evidence. The Court used this as the baseline for evaluating whether the BIA’s reversal could stand.
  • Myrie v. Att'y Gen., 855 F.3d 509 (3d Cir. 2017)
    Central to the standard-of-review dispute. Myrie reiterates the two-prong CAT analysis and allocates standards of review: prong one (“what is likely to happen”) is factual and reviewed for clear error; prong two (whether predicted harm is legally “torture”) is reviewed de novo. The Court treated the BIA’s reversal as a permissible de novo legal determination under prong two, not as impermissible factual reweighing.
  • Kaplun v. Att'y Gen., 602 F.3d 260 (3d Cir. 2010)
    Quoted through Myrie for the two-step CAT inquiry and the fact/law split. It undergirds the Court’s conclusion that the BIA may accept an IJ’s forecast of detention yet still reverse if detention does not, on this record, establish a likelihood of “torture” as a legal matter.
  • Chavarria v. Gonzalez, 446 F.3d 508 (3d Cir. 2006)
    Cited for the “compelled to conclude otherwise” gloss on substantial evidence review. The Court relied on it to reject petitioner’s claim that the record required reinstating CAT deferral.
  • Hernandez Garmendia v. Att'y Gen., 28 F.4th 476 (3d Cir. 2022)
    Used to emphasize that generalized evidence of torture in a country is insufficient without specific evidence that the applicant is “more likely than not to be singled out.” The Court treated this as directly supporting the BIA’s rationale that Attoev’s evidence showed, at most, a general possibility of abuse, not an individualized likelihood of torture.
  • Denis v. Att'y Gen., 633 F.3d 201 (3d Cir. 2011)
    Cited via Hernandez Garmendia for the same individualized-risk principle. It bolstered the Court’s view that petitioner’s record lacked the specificity needed to meet the CAT “more likely than not” burden.
  • Herrow v. Att'y Gen., 93 F.4th 107 (3d Cir. 2024)
    Cited as a contrast: where evidence shows extremely high torture rates for a defined group (e.g., “more than fifty percent” of a group), it may support a clear probability of torture. The Court used it to illustrate what kind of quantitative or group-based evidence might satisfy the burden—implicitly highlighting the comparative weakness/ambiguity of Attoev’s reports.
  • Kang v. Att'y Gen., 611 F.3d 157 (3d Cir. 2010)
    Cited for the proposition that reversal is warranted when the agency “overlook[ed] evidence.” The Court invoked it to conclude the BIA did not ignore the record evidence, including the ECHR decision and country reports.

B. Legal Reasoning

1. The governing CAT framework and burdens

The Court began from the regulatory structure: a person ineligible for CAT withholding because of danger to national security may still pursue CAT deferral. 8 C.F.R. § 1208.17(a). The applicant bears the burden. 8 C.F.R. § 1208.17(d)(3). Under the Third Circuit’s longstanding approach (as described in Myrie quoting Kaplun), the adjudicator must decide:

  1. what is likely to happen to the applicant if removed; and
  2. whether what is likely to happen meets CAT’s legal definition of “torture.”

The opinion emphasized the standards of review attached to each step: “likely to happen” is a factual forecast (clear-error review), while the “torture” determination is a legal question (de novo review). “Torture” is defined as an “extreme form of cruel and inhuman treatment” and excludes “lesser forms” of cruel, inhuman, or degrading treatment. 8 C.F.R. § 208.18(a)(2).

2. Why the BIA did not engage in improper de novo fact review

Attoev’s primary procedural challenge was that the BIA supposedly reweighed evidence and substituted its judgment for the IJ’s. The Third Circuit rejected that characterization by pointing to what the BIA actually did:

  • The BIA accepted the IJ’s adverse credibility finding.
  • The BIA accepted the IJ’s factual finding that Attoev was “more likely than not” to be detained upon removal, expressly concluding that this finding was “not clearly erroneous.”
  • The BIA reversed because the IJ “erroneously conflate[d]” the risk of arrest/detention with the risk of torture—i.e., the BIA treated the IJ’s ultimate inference (detention implies torture) as legally insufficient under the CAT standard.

In other words, the Third Circuit framed the BIA’s action as a permissible legal correction: even if detention is likely, the applicant must still prove it is more likely than not that he will be tortured, and generalized evidence of torture in the country does not bridge that gap without an individualized showing.

3. Treatment of the ECHR decision and the “ill-treatment” vs. “torture” distinction

A key evidentiary point was the petitioner’s reliance on an ECHR ruling stating he faced a “real risk of ill-treatment” if removed. The Third Circuit held the BIA did not “disregard” that ruling; rather, it treated it as probative of “ill-treatment,” which the BIA correctly explained is not automatically equivalent to “torture as defined by the CAT.”

The Court’s reasoning here reinforces a doctrinal boundary: international or foreign tribunal findings describing “ill-treatment” may be relevant but do not, without more, satisfy the CAT’s more demanding definition. This is especially true where the domestic record contains an adverse credibility determination and lacks evidence that the applicant would be specifically targeted for torture.

4. Individualization requirement: generalized conditions are not enough

On the merits and substantial evidence review, the Court relied heavily on the principle articulated in Hernandez Garmendia v. Att'y Gen.: generalized statements about torture in a country are insufficient; the applicant must provide specific evidence showing he is more likely than not to be singled out.

The Court applied this to the country-conditions materials and anecdotal incidents. It highlighted an internal tension in petitioner’s theory: much evidence suggested Uzbek authorities use “trumped up” charges against religious individuals, but the IJ had already found there was no evidence the Uzbek government fabricated charges against Attoev as a pretext for religious persecution and that Attoev failed to explain why he was being targeted beyond allegations of terrorism. With credibility rejected and a pretext theory unsupported, the record did not compel the conclusion that detention would more likely than not entail torture.

5. Why the record did not compel reversal (substantial evidence)

The Court held substantial evidence supported the BIA because:

  • The petitioner’s testimony was found not credible (a finding the BIA and Court left intact), shrinking the individualized evidentiary basis for CAT protection.
  • The remaining evidence showed torture exists in Uzbekistan, but did not show a clear probability that this petitioner would be tortured, as opposed to suffering non-CAT-level “ill-treatment” or facing uncertain treatment.
  • The Commission Report’s own caution that its “quantitative and statistical conclusions ... are by no means definitive,” combined with the fact that fewer than half of profiled individuals alleged torture (as characterized by the Court), undercut any claim that the report compelled a finding of likely torture.
  • The State Department Reports discussed both “torture and ill-treatment” and described torture as “commonplace” in pretrial detention, but did not provide a basis to conclude that Attoev, in particular, would more likely than not be tortured.

Applying Chavarria v. Gonzalez, the Court concluded a reasonable factfinder would not be compelled to reach the opposite conclusion.

C. Impact

Although designated “NOT PRECEDENTIAL,” the decision illustrates and reinforces several operational lessons for CAT deferral litigation in the Third Circuit:

  • Arrest/detention is not enough. Even if removal will likely result in arrest or detention, CAT relief still requires proof that the applicant is more likely than not to suffer harm meeting the regulatory definition of torture.
  • “Ill-treatment” is a lower category than CAT “torture.” Evidence—whether from foreign tribunals or reports—framed as “ill-treatment” may not satisfy CAT’s “extreme” threshold without detail showing severity, intent, and likelihood.
  • Individualized risk is crucial, especially after an adverse credibility finding. Where an IJ makes an adverse credibility determination, country reports and generalized evidence must do far more work; applicants need a clear, individualized pathway connecting their circumstances to likely torture.
  • Clarification of BIA review boundaries. The decision provides a template for distinguishing improper BIA fact reweighing from permissible de novo legal review under the second prong of the Kaplun/Myrie framework.

IV. Complex Concepts Simplified

CAT deferral vs. CAT withholding
Both are forms of protection under CAT. Withholding is stronger but can be barred for certain reasons (e.g., danger to national security). Deferral is more limited and may still be available even when withholding is barred, but it still requires proving likely torture.
“More likely than not”
A probability standard—roughly meaning greater than 50% likelihood that the applicant will be tortured if removed.
The two-prong CAT inquiry
(1) Fact prediction: what will happen to the person if returned (e.g., detained, interrogated). (2) Legal classification: whether the predicted harm qualifies as “torture” under the regulations.
Clear error vs. de novo review
“Clear error” is deferential: the BIA must accept the IJ’s fact findings unless they are plainly wrong. “De novo” is non-deferential: the BIA may decide the legal issue anew, such as whether predicted harm legally qualifies as “torture.”
Generalized vs. individualized evidence
Generalized evidence shows that torture occurs in a country. Individualized evidence links the applicant personally to a specific risk—showing why this person is more likely than not to be tortured.

V. Conclusion

The Third Circuit’s decision affirms that CAT deferral demands more than proof of likely detention and more than generalized evidence that torture occurs in the country of removal. Under the Myrie/Kaplun framework, an applicant must connect the predicted post-removal events to a legally sufficient showing that torture—as an “extreme” form of mistreatment—is more likely than not to occur to him personally. The BIA acted within its authority by accepting the IJ’s factual prediction of detention yet reversing on the legal insufficiency of the torture showing, and substantial evidence supported that outcome in light of the adverse credibility finding and the largely generalized nature of the supporting reports.