Immigration courts' caseload surged 150%. Video hearings lagged

GAO⧉ -- Congress's independent, nonpartisan auditor -- spent this review inside the Department of Justice's Executive Office for Immigration Review (EOIR), which runs 73 immigration courts and three video-only adjudication centers with 634 judges deciding whether noncitizens facing removal can stay in the country.▸ EOIR's own officials and the judges, prosecutors, and defense attorneys GAO interviewed describe remote hearings -- held over WebEx, older video-teleconference (VTC) links, or by telephone, as opposed to everyone appearing in the same room -- as a genuine convenience: less travel, more flexibility, faster dockets. The hearing data GAO pulled from EOIR's own case-tracking system tells a narrower story. As the courts' total caseload grew almost 150 percent from fiscal year 2022 through 2025, remote hearings grew only about 50 percent -- while in-person hearings grew almost 250 percent, five times faster. Video didn't absorb the surge. In-person courtrooms did.

From fiscal year 2022 through 2025, EOIR held nearly 6 million hearings -- about 63 percent (3.78 million) in a courtroom, the remaining 37 percent (2.18 million) by remote medium.▸ Of those remote hearings, 78 percent used WebEx, the internet video platform EOIR adopted during the pandemic; the rest used older courtroom-to-courtroom video links or plain telephone. That 37 percent figure sounds like meaningful adoption of remote technology, until it's read against the growth numbers: the docket itself nearly doubled and a half again, and remote hearings' share of it shrank rather than grew. EOIR officials told GAO the shift traces to post-pandemic normalization -- as COVID-era restrictions eased, courts scheduled more hearings overall, and more of the added volume landed back in person.

The pattern is not uniform across the docket.▸ Detained respondents -- the roughly 731,000 hearings, 12 percent of the total, for people held in ICE custody -- ran 73 percent remote, because it costs less to connect a detention facility by video than to transport someone to a courthouse. It's the opposite story for master calendar hearings, a respondent's very first appearance, where a judge confirms the person understands the charges against them and the proceedings ahead: those 4.9 million hearings, the largest single category, ran only about one-third remote. Judges told GAO they generally want that first encounter in person, to verify identity and comprehension directly.

Here is the mechanism GAO's report documents, plainly, that explains part of why in-person hearings keep winning: current WebEx technology cannot provide simultaneous interpretation -- the real-time, headphones-based translation that lets a non-English-speaking respondent follow a hearing as it happens -- whenever the respondent and the interpreter aren't physically together.▸ In a courtroom, that's solvable; over WebEx, it currently is not.

The fallback is consecutive interpretation, where every statement pauses so the interpreter can translate it afterward, which EOIR officials and court stakeholders told GAO can take double or triple the time of a hearing with simultaneous interpretation.

That gap matters at scale, because English-only hearings are the exception in this system, not the rule: about 5.7 million of the 5.96 million hearings GAO examined -- roughly 96 percent -- involved a respondent speaking one of almost 600 languages other than English, Spanish alone accounting for 76 percent of the total.▸ GAO's report does not draw a straight line from the interpretation gap to the growth numbers -- it presents both as separate findings, not one causing the other -- but the two sit next to each other in the same document: the tool that's supposed to make hearings more efficient works worst for exactly the population that makes up almost the entire docket.

The clearest data point on that question sits in GAO's own appendix tables, not its main findings: among the ten most common languages spoken in immigration court, English speakers had their hearings held remotely 51 percent of the time -- well above the system-wide 37 percent average, and higher than every language on the list except Mandarin (60 percent), Punjabi (57 percent), and Turkish (52 percent), each spoken by a far smaller number of respondents than English or Spanish.▸ Spanish speakers, the largest group by far at 76 percent of all hearings, sat at 33 percent remote, below the system-wide average. Nothing in the report says the interpretation limitation causes that 18-point gap between English and Spanish speakers -- other factors, like which courts and case types Spanish-speaking respondents' hearings are concentrated in, could explain some or all of it, and GAO didn't test for a link. What GAO does confirm is that the technology itself treats English and non-English hearings differently, and that no one has checked whether that difference shows up in who gets a fair, efficient hearing and who doesn't.

That last omission has a paper trail.▸ GAO's 2017 review of the same system made four recommendations about remote-hearing technology; EOIR implemented three. The fourth -- directing EOIR to study what effect, if any, remote versus in-person hearings have on case outcomes -- remains open. EOIR's answer to GAO in this report is that it checked court appeals for complaints about hearing medium and found fewer than 1 percent citing the issue in 2019 and 2020, which EOIR treats as evidence the medium doesn't affect outcomes. That's a self-conducted check, from years before this report's FY2022-2025 study window and before EOIR itself had documented the interpretation limitation described above, and this report does not revisit it independently.

This report also carries no new recommendations of its own: the entire document is a description of what happened and what stakeholders think of it, not a set of directives anyone at DOJ or EOIR is now on the hook to act on.▸ Separately, EOIR rescinded its 2022 guidance on when judges should grant remote-hearing requests in March 2025, calling it "unhelpful" and an infringement on judicial independence -- leaving a 2020 policy in place that gives judges no specific criteria for the decision at all.