evidoria

← Back to browse

Good practice

PACC — Austria's AI-driven tax fraud and evasion detection

Austria · Vienna · See the Austria profile

Austria's Predictive Analytics Competence Centre (PACC) uses machine learning to screen millions of tax cases for fraud. In 2023 it analysed 6.5M cases, recovering €185M; in 2024, it screened 6.6M tax cases and 23.4M compliance records, yielding €354M.

€185 million
Tax revenue recovered (2023)
€354 million
Tax revenue recovered (2024)
6.5 million
Tax cases screened (2023)
6.6 million
Tax cases screened (2024)
27.5 million
Compliance cases assessed (2023)
23.4 million
Compliance records assessed (2024)

Details

Maturity
Established
Promoter
Austrian Federal Ministry of Finance (BMF) — Predictive Analytics Competence Centre (PACC)
Period
2016-present
Keywords
tax fraud detection, VAT fraud, predictive analytics, machine learning, supervised learning, text mining, revenue compliance

Context

PACC is a special unit within Austria's Federal Ministry of Finance, established 2016–17, that uses supervised machine learning and text-mining models across four analytical teams (Predictive Analytics, Advanced Analytics, Tax Analytics, Customs Analytics) to risk-score income tax, corporation tax, VAT and customs cases; risk scores inform human audit teams, who retain final decision-making authority.

Results

In 2023, PACC's models screened approximately 6.5 million tax cases and 27.5 million compliance cases, resulting in approximately €185 million in recovered tax revenues. In 2024, 6.6 million tax cases and 23.4 million compliance records were assessed, yielding €354 million in additional tax revenue.

Implementation

Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful