Congress ordered a count of border deaths. The report came up short
Summary
The Missing Persons and Unidentified Remains Act tells CBP exactly what its annual report must contain — including how often each rescue beacon was activated by someone in distress. GAO's review of the newest report, out today: three of the four beacon-data elements GAO checked are missing, two have never appeared in any of the four reports filed, and information earlier reports carried was removed as 'not specifically required.'
The documents
Five primary documents, read against each other: today's GAO-26-108818⧉; the statute⧉ whose section 5(b) is the checklist; 's April 2025⧉ and April 2022⧉ reviews of the same program; and 's public Border Rescues and Mortality Data page⧉, sealed today. The 2022 review is where this story starts: found Border Patrol 'has not collected and recorded, or reported to Congress, complete data on migrant deaths, or disclosed associated data limitations.' Border Patrol accepted all three recommendations and fixed the data collection. Four years later, the completeness problem has migrated from the database to the report.
The stakes
The program these reports describe is Border Patrol's rescue-and-death-reduction effort, running since 2017 across a 45-county target zone and, since fiscal 2023, one northern sector. In fiscal 2024 it recorded 3,302 rescue events and 5,679 people rescued⧉; rescue-beacon activations were linked to 70 of those rescues. Whether 70 is a good return on the beacon network is precisely the kind of question Congress's required elements would let anyone answer — activations per beacon, distress types, beacons per sector. Those numbers have never been public. The effectiveness of towers built to keep people alive in the desert cannot be computed from the public record — which is the finding.
View data as table
| People rescued | 5,679 | |
|---|---|---|
| Rescue events | 3,302 | |
| Rescues linked to beacon activations | 70 | the beacons whose data goes unreported |
The disagreement between documents
The disagreement is between a statute and four reports. Section 5(b) is not ambiguous — quotes it, Border Patrol officials, shown the analysis, agreed the act requires the omitted information⧉. The explanations on the record: officials 'inadvertently omitted' the beacon count and were 'not aware' the rest was required; the internal operating procedures that assign the report to the manager never list what the report must contain. Meanwhile the report moved in the other direction from completeness: the fiscal 2025 edition removed the rescue-incident counts and types, placard counts, and placard-linked rescues that earlier editions carried — the official reason being that the act didn't specifically demand them. A reporting statute produced a report that contains less than the reports filed before anyone checked.
View data as table
View data as table
| Rescue incidents by number and type | 1 | exposure to heat, water-related, etc. |
|---|---|---|
| Rescue placards deployed | 1 | |
| Rescues associated with placards | 1 |
What happens next
's single recommendation is procedural on its face — write the statutory checklist into the operating procedures — and concurred. Two dated items follow: Border Patrol expects to complete its first full program assessment by September 2026⧉, under the evaluation design 's 2025 report demanded; and the next annual report the act requires, will show whether the checklist finally makes it into the document. The desk will read both against section 5(b), which does not change — including its fifth, never-reviewed element: a stakeholder-consulted assessment of whether more beacons are needed.
- 's newest report under the Missing Persons and Unidentified Remains Act omits 3 of the 4 rescue-beacon elements the statute requires; beacon activation counts and distress types have never been reported in any of the four reports filed.
- The FY2025 report also removed information earlier reports included — rescue counts by type, placard numbers, placard-linked rescues — as 'not specifically required.'
- The program at stake rescued 5,679 people in FY2024; whether its beacon network works cannot be computed from the public record.
- 's 2022 finding was incomplete death data; the 2026 finding is incomplete reports. concurred with the fix; the September 2026 program assessment and the next annual report are the tests.
Method notes. All figures are Border Patrol data as reported by ; the ~1.2% beacon share divides beacon-linked rescues by all people rescued and says nothing about beacon effectiveness — the activation denominators needed for that have never been published, which is the point. The 2022-to-2026 continuity is about reporting discipline: the 2022 recommendations on data collection were implemented, and today's finding concerns the statutory report's completeness. The FY2025 report's location map partially satisfies the location element; 's count of omissions (number per sector, activation frequency, distress types) is used as published. 's public statistics page is sealed as of today as the public counterpart of the congressional reports.
Sources(5) ▾
- U.S. Government Accountability Office, U.S. Customs and Border Protection: Actions Needed to Ensure Complete Reports to Congress on Unidentified Remains and Rescue Beacons (GAO-26-108818) (2026-07-14) — Released today. 's fiscal year 2025 report to Congress omitted required rescue-beacon information — the number deployed per sector, activation frequencies, and distress types — and 's analysis found activation frequencies and distress types have never appeared in any of 's reports. also removed information prior reports had carried (rescue incident counts and types, placard counts, placard-linked rescues) because it 'was not specifically required.' Also: FY2024's 3,302 rescue events and 5,679 people rescued; 70 beacon-linked rescues; concurrence with the one recommendation; a program assessment expected by September 2026. gao.gov · original document
- U.S. Government Accountability Office, Border Security: Additional Actions Needed to Evaluate the Missing Migrant Program (GAO-25-107548) (2025-04-07) — The April 2025 predecessor: recommended Border Patrol develop an evaluation design and collect activity information clearly linked to the program's scope and purpose — the recommendations whose implementation the 2026 report tracks. gao.gov · original document
- U.S. Government Accountability Office, Southwest Border: CBP Should Improve Data Collection, Reporting, and Evaluation for the Missing Migrant Program (GAO-22-105053) (2022-04-20) — The 2022 finding this story descends from: 'Border Patrol has not collected and recorded, or reported to Congress, complete data on migrant deaths, or disclosed associated data limitations.' Border Patrol agreed with all three of its recommendations and later implemented them. gao.gov · original document
- U.S. Congress / GPO, Missing Persons and Unidentified Remains Act of 2019 (Public Law 116-277) (2020-12-31) — The reporting law. Section 5(b) requires 's annual report to include, for rescue beacons: the number in each Border Patrol sector, the specific location of each beacon, the frequency with which each was activated by a person in distress, and a description of the distress, if determinable. govinfo.gov · original document
- U.S. Customs and Border Protection, CBP: Border Rescues and Mortality Data (public statistics page) (2026-07-14) — 's public-facing rescues-and-mortality statistics page, sealed as of today — the public counterpart of the congressional reports at issue. cbp.gov · original document
Comments
Always open. Logged-in readers can annotate paragraphs in place.
In 2020, Congress passed the Missing Persons and Unidentified Remains Act⧉ and specified, element by element, what must tell it every year: the unidentified remains found along the border, and — for the rescue beacons meant to keep people from becoming remains — the number in each sector, each beacon's location, how often each was activated by a person in distress, and what kind of distress. GAO's review⧉ of 's newest report, released today, finds three of the four beacon-data elements it checked missing. Two of them — activation frequency and distress type — have never appeared in any of the four reports has filed. And the fiscal 2025 report also dropped information earlier reports had volunteered, because it 'was not specifically required.'