Cross-Referencing Every Taxpayer — Malta's SAS-Powered AI Push Against VAT and Tax Evasion
Malta
Malta committed €3m to deploy SAS AI tools that cross-reference registries and bank data to flag VAT fraud; officials attribute …
Hungary · Budapest · See the Hungary profile · See the Budapest profile
Evidence: Observational / pre–post Top 24% 73/100 · Ask Evidence Copilot about this practice
Hungary requires all businesses to report invoices to NAV in real-time XML since 2021; ML risk-scoring targets suspicious companies before returns are filed. Its VAT compliance gap fell from 21.7% (2013) to 2.3% (2022)—best in the EU—matching each phase of the mandate rollout.
Hungary's National Tax and Customs Administration (NAV) has required real-time XML invoice reporting to its Online Számla platform since 2018, expanded to universal mandatory coverage of all B2B, B2C, export and intra-community invoices from January 2021; an ML risk-scoring working group (MIMCS) was formalised in 2022 to flag high-risk businesses for NAV's 132,000+ annual inspections.
Hungary's VAT compliance gap fell from 21.67% in 2013 to 2.31% in 2022 - the best result in the EU - with each phased expansion of the reporting mandate matched by a measurable gap reduction, per successive European Commission VAT Gap Reports.
Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.
National / regional programmes Digital Europe Programme
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.
Malta
Malta committed €3m to deploy SAS AI tools that cross-reference registries and bank data to flag VAT fraud; officials attribute …
Slovenia
Slovenia's FURS runs two ML risk-scoring models in its eDIS platform — one flags irregular VAT-refund claims since 2019, another …
Rwanda
A study built with Rwanda Revenue Authority data used MiniLM, UMAP and HDBSCAN to harmonize 4.1 million multilingual product names …
United States of America
The IRS grew from 10 to 126 active AI use cases between 2022 and mid-2025 to flag high-risk tax returns …
Open full copilot Grounded in cited practices — always check the sources.