A CBP anomaly detection pilot at land ports of entry (H.R.9566) — limited to the Arizona border, with staffing shortages among the things evaluated
A bill requiring U.S. Customs and Border Protection to implement, within two years of enactment, a pilot program using an AI anomaly detection algorithm at land ports of entry along the Arizona border. The matters assessed include not only detection accuracy and processing times but inefficiencies and understaffing among officers.
Bill overview (primary data)
- Bill numberH.R. 9566
- TypeHouse Bill
- Congress119th Congress
- Latest actionReferred to the Subcommittee on Border Security and Enforcement.（2026-07-07）
Key points
- A bill requiring CBP to implement, within two years of enactment, a pilot using an AI anomaly detection algorithm at land ports of entry along the Arizona border.
- The purpose is improving screening, efficiency, inspection and interdiction of illicit drugs and contraband, positioned as assisting CBP officers.
- The first assessment item is assisting with inefficiencies and understaffing of officers, not detection accuracy.
- Improving thoroughness of inspections and improving processing times appear in the same list, two things that usually pull against each other.
- It is designed as a pilot narrowed to one border rather than a request for full deployment.
- Anomaly detection may catch unanticipated methods while also flagging the merely unusual, so it is positioned to narrow candidates.
1Trying it at one border
The scope of this bill is land ports of entry along the Arizona border, not ports nationwide. It is designed as a pilot with the range narrowed, and does not ask for full deployment. Where the effect of a technology depends on local conditions, the order taken is to try it within a limited range and gather data before widening.
A deadline of two years to begin is also set, drafting that avoids a state of continued study with nothing moving.
2Understaffing sits among the evaluation items
The first item listed for assessment is neither accuracy nor processing time but assisting with inefficiencies and understaffing of officers. Detection performing better and a limited number of people being able to keep up are different kinds of value, and the second bears directly on operations.
Improving thoroughness and improving processing times appearing in the same list is notable as well, because those two usually pull against each other. Trying to measure both at once is where the constraint the field is under shows through.
3Why anomaly detection
The method chosen is an anomaly detection algorithm. Rather than being taught in advance what is illicit, it finds what departs from the ordinary — which can respond to methods nobody anticipated while also flagging things that are merely unusual. That the text repeats the purpose of assisting CBP officers reads as positioning it to narrow candidates rather than to replace judgment.
The Secretary of Homeland Security is separately directed to ensure that its use satisfies departmental requirements.
4Not teaching it, but finding what differs
The algorithm used is anomaly detection. What it is good and bad at differs from a method taught in advance what is unlawful.
That the text repeats "to assist inspection by CBP officers" reads as positioning it to narrow candidates rather than replace judgement. The scope is confined to land ports of entry on the Arizona border rather than nationwide, with a two-year deadline from enactment to begin. The first evaluation criterion listed is not accuracy or processing time but addressing officer inefficiency and staffing shortages.
Why it matters
The value of deploying AI is not always captured by accuracy metrics. This bill puts relief of understaffing at the head of what is assessed, which points to a setting where what is wanted is throughput more than detection power. Starting from a narrowed pilot is a useful pattern for advancing adoption decisions in uses whose effect depends on conditions.
FAQ
What is an anomaly detection algorithm?
Why only the Arizona border?
Would the AI replace judgment?
Sources (primary)
Source: Congress.gov (Library of Congress; U.S. legislative materials, public domain). Links go to the official site.
- Congress.gov (bill page, original)
- H.R. 9566（119th Congress）