Catalogue: AI in the Public Sector clear
B’Odogwu (Nigeria’s Unified Customs Management System) applies AI risk-profiling and ML pattern detection across 34 commands. Between Oct 2024 and May 2025 …
Hungary requires all businesses to report invoices to NAV in real-time XML since 2021; ML risk-scoring targets suspicious companies before returns are …
Romania's ANI automatically cross-references procurement officials' identities with registered bidders before contracts are signed; over 100,000 procedures worth RON 72 billion screened, …
NECTEC's AI chatbot (LINE) classifies and routes citizen complaints nationally. June 2022–July 2024: 947,769 complaints handled, 77% resolution rate, resolution time cut …
Costa Rica’s tax directorate uses the open-source CIAT e-IAD ML model to detect anomalies in e-invoices. By 2026 it exposed a ~₡80 …
Kazakhstan's Digital Family Card uses predictive scoring across 100+ socio-economic indicators from 20+ agencies to proactively identify and serve vulnerable families. Since …
STF's Victor AI classifies Extraordinary Appeals by 'general repercussion' theme in ~5 seconds vs 44 minutes manually, handling ~50,000 cases/year. Built with …
Chile's RSH algorithm classifies 8.99 million households (85.6 % of the population) into socioeconomic tiers that gate access to all means-tested state …
VDAB has deployed multiple AI systems since 2018: a multilingual deep-learning job-matching engine; Orient 2.0 (orientation test reduced from 45 to 10 …
Austria's Predictive Analytics Competence Centre (PACC) uses machine learning to screen millions of tax cases for fraud. In 2023 it analysed 6.5M …
New Zealand's Inland Revenue uses ML to screen 3+ million tax returns for compliance risk. In H1 2024 this drove a 50% …
South Korea's Clova CareCall places AI-driven phone check-ins with ~50,000 isolated elderly across 150+ institutions. Peer-reviewed studies show depression improvements and a …
Canada Revenue Agency uses ML and AI-powered analytics to detect unwarranted GST/HST refund claims. Since 2017–18, over C$1.1 billion in carousel-scheme fraud …
40 state, municipal and regional AI pilots (2020-22) were independently evaluated in 2024: ~15% in operation, 45% ongoing; data quality and legal …
Building on SAMAI 1, which certified 1.1 million Saudi citizens in AI during 2025 with 52% female participation, SDAIA launched SAMAI 2 …
Kazakhstan's central bank added Ridge, Lasso and Elastic-Net machine learning to its inflation forecasting system, cutting pseudo-out-of-sample RMSE 15-19% across three test …
The National Bank of Ukraine published a peer-reviewed method that counts vehicles in satellite images to nowcast Kyiv's wartime economy, building on …
Banco de México built five real-time GDP-nowcasting models and back-tested them over 12 quarters (2014 Q2-2017 Q1); the winning bridge-equation average beat …
San José's City Clerk replaced budgeted human interpreters with Wordly AI translation for council meetings, cutting annual costs from $400,000 to $82,000, …
A D.R. Congo provincial tax authority used drone imagery and machine learning to build a property tax roll for Kananga's 106,000+ buildings, …
Hongkong Post's 15 NEC-built mail-sorting systems, running at the Central Mail Centre since 2008, were upgraded in 2017 to read handwritten and …
Uzbekistan's Tax Committee, with IMF support, built a real-time AI risk-scorer for mandatory e-invoices, red-flagging suspected fraud; its existing analytics already found …
Belgium's Youth for Climate movement used CitizenLab's NLP platform to collect 1,700 ideas and 32,000 votes, auto-clustering them into 15 priorities handed …
Canada's federal Red Tape Reduction Office paired BizPal's AI-generated plain-language rule summaries, and a parallel AI regulation-scanner piloted by Ontario, with a …
Open full copilot Grounded in cited practices — always check the sources.
