G-M-I-: Why the Board Rejected an Expert’s Prediction of Torture

If you fear being tortured after removal, an expert’s report may help explain a danger you cannot document on your own. The report needs to connect its sources to your circumstances. In this case, the Board found that a key example concerned a different kind of prosecution, while a striking statistic answered a different question.

A Chinese citizen sought protection from removal to China under the Convention Against Torture, known as CAT. An immigration judge granted deferral of removal, relying on an expert’s prediction of torture. The government appealed. In Matter of G-M-I-, 29 I&N Dec. 431 (BIA 2026), the Board of Immigration Appeals sustained the appeal and vacated that protection. Its February 4, 2026 decision explains why the expert’s factual foundation mattered to the result.

The expert predicted what would happen after return

The respondent had been convicted in 2014 of conspiring to import at least 150 kilograms of cocaine into the United States and sentenced to 14 years in prison. He conceded removability and sought CAT deferral. After an earlier Board remand, the judge granted that protection on March 26, 2025.

The expert specialized in China’s legal system. Through two affidavits and testimony, he predicted that China would learn of the conviction and subject the respondent to torture or execution. He relied on accounts of harsh drug penalties, concern for China’s reputation, use of the death penalty and the government’s ability to monitor people. The Board’s account of the expert evidence shows that the judge gave his statements and testimony full weight.

The appeal tested the connection between those materials and the predicted treatment of someone returning after a conviction abroad. That connection became especially important when the Board checked the examples the expert offered.

The Canadian example described a different prosecution

The expert described a Canadian national arrested on entering China after a drug conviction in Canada. The Board read the article submitted in support of that account and found a different sequence: The Canadian citizen had been arrested and convicted in a Chinese court for trafficking drugs in China.

According to the Board’s explanation of the article, the man initially received a 15-year sentence. A death sentence followed a retrial. The article also alleged a political connection to Canada’s arrest of a Chinese telecommunications executive. The Board found that this example had little relevance to the respondent’s circumstances.

The location of the prosecution was central to the comparison. Evidence about punishment for conduct prosecuted in China addressed a different situation from the respondent’s conviction in the United States and anticipated return. Checking the cited article exposed that mismatch.

What the 95 percent figure actually counted

The expert also cited a law-journal statement that 95 percent of death sentences in China involved three categories together: Drug crimes, intentional homicide and robbery. As the Board explained when examining the statistic, the figure did not establish how likely a person convicted of a drug crime was to receive a death sentence, especially someone convicted in another country.

The distinction is between starting with death sentences and asking which crimes they involved, and starting with people convicted of a particular crime and asking what happened to them. The reported figure described the first group. The expert’s prediction concerned the second question. The Board found that the source did not supply that missing probability.

The Board also identified gaps the expert acknowledged: Little reliable information about torture and executions, and little official information about the handling of drug traffickers returned from abroad. Those gaps mattered because the report used the available material to predict this respondent’s treatment.

The Hong Kong example needed an explained connection

The expert’s other example concerned a tycoon abducted from a Hong Kong hotel room in 2017 and taken into extralegal detention. The Board found too little context explaining what that person and the respondent had in common. The example’s seriousness supplied no explanation of why it supported the particular prediction being made.

In its assessment of the expert’s factual foundation, the Board brought these problems together. The expert relied on two examples that did not provide a sound comparison to the respondent’s situation. It found clear error in the judge’s assignment of significant weight without reasonably addressing those deficiencies.

The Board separated the evidence review from the CAT burden

The decision used two review standards. The Board reviewed the judge’s predictions about future events for clear error. It reviewed the ultimate legal determination of whether the CAT burden was met afresh, or de novo. The distinction tracks the Board’s review rules, which separate factual findings from legal questions.

After addressing the weight given the expert, the Board examined whether the evidence established the required likelihood of torture. It assumed for this analysis that Chinese authorities would learn of the conviction and detain the respondent on arrival. Even on that assumption, it found the evidence inadequate to show that he would more likely than not be tortured.

That is where the defective comparison and statistic affected the protection claim. The Board found insufficient evidence about the treatment of people convicted of trafficking outside China, or a particular government interest in this respondent. It concluded that the expert’s endorsement of the same generalized material did not establish the required individualized probability.

The order vacated the grant of protection

The Board’s order sustained the government’s appeal and vacated the March 2025 CAT deferral. It stated that removal would proceed under the immigration judge’s removal order. The operative result was the loss of the deferral the judge had granted.

For the broader evidence framework, section 1208.16(c) requires consideration of all relevant evidence about future torture, including past torture and country conditions. The useful contribution of G-M-I- is its close examination of what particular sources supported. Our guide to connecting past harm to the risk of return addresses that different evidentiary task.

Read the source alongside the prediction

When reviewing an expert report with counsel, place the important prediction beside the source offered to support it. For an example involving a conviction, identify where the conduct occurred, where it was prosecuted and what happened on return. For a statistic, identify the people or events counted and the conclusion the number actually supports.

Then compare that explanation with the judge’s findings and any government objection. In this case, the location of the Canadian prosecution had already been raised by the government. Addressing a conflict in the cited material matters to the judge’s explanation of why an inference is persuasive. Bring the full expert report, its supporting articles, the conviction record and the relevant decisions so counsel can evaluate those connections.

Sources

The official G-M-I- opinion supplies the facts, the Board’s account of the expert’s sources, its reasoning and its order. The linked regulations supply the current review and CAT evidence framework. The Canadian example and journal statistic here are described as the Board examined them in the submitted record.