AI Risk Screening for Customs — Tunisia's National Selectivity System Targets Fraud
Tunisia
In May 2026 Tunisia's Customs Authority added a machine-learning module to its selectivity system to flag high-risk shipments by origin, …
Palau · Koror · See the Palau profile · See the Koror profile
Evidence: Observational / pre–post Top 24% 73/100 · Ask Evidence Copilot about this practice
Palau's Bureau of Customs and Border Protection deployed UNCTAD's ASYCUDAWorld 4.4 with an AI "dynamic selectivity" risk engine in January 2024. Within months, revenue rose 20% year-on-year and clearance times fell from a day to under two hours.
Palau, one of 15 Pacific nations covered by the EU-funded IMPACT project, switched from manual/legacy customs processing to UNCTAD's ASYCUDAWorld platform on 22 January 2024, run by the Bureau of Customs and Border Protection (BCBP) under the Ministry of Finance.
ASYCUDAWorld 4.4 ships a "dynamic selectivity" module — a self-improving machine-learning risk model that scores each customs declaration for fraud/non-compliance risk using historical declaration data, and automatically updates its own risk profiles from enforcement outcomes, a feedback loop UNCTAD describes as a machine-learning model.
BCBP's own comparison shows customs revenue up 20% from January to August 2024 versus the same period in 2023; officials also report that goods clearance, once requiring roughly a full day, now typically takes one to two hours, with 297 registered ASYCUDAWorld users. A separate UN briefing for the Federated States of Micronesia's 2024 launch cited an even sharper 53% revenue jump in Palau's first two months live, suggesting an early-adoption spike that settled into the sustained 20% annual gain.
Public reporting to date comes from UNCTAD/EU project communications and Palauan press rather than an independent third-party audit, and no disaggregated data on the AI risk engine's false-positive/negative rates has been published.
National / regional programmes
Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.
Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.
Where this practice's information was retrieved from, and when.
Documented replications and adaptations of this practice elsewhere.
| Place | When | Adopter | Outcome |
|---|---|---|---|
| Micronesia | 2024 | Federated States of Micronesia (customs authority) | Launched the same ASYCUDAWorld dynamic-selectivity AI customs risk engine under the EU-funded IMPACT project. |
Tunisia
In May 2026 Tunisia's Customs Authority added a machine-learning module to its selectivity system to flag high-risk shipments by origin, …
Iraq
Iraq's Customs Authority deployed UNCTAD's ASYCUDA World, whose newest generation adds AI-based risk-scoring to sort shipments into inspection channels. Revenue …
Mexico
Mexico's tax authority uses AI to flag fraudulent e-invoicing and shell companies, and runs OrientaSAT, a chatbot serving taxpayers — …
Algeria
Algeria's customs authority launched ALCES, a clearance platform whose risk module uses AI to flag import/export fraud risk. It processed …
Open full copilot Grounded in cited practices — always check the sources.