Italy Leverages AI to Streamline VAT Recovery Process
Italy’s Agenzia delle Entrate is set to automate VAT refund processing with artificial intelligence by 2026. Early trials on low-risk VAT claims have shown promising results, sparking further investment in AI-driven solutions. This development aims to improve refund speed while enhancing fraud detection, with the final decisions still being made by humans.
AI-powered chatbots have already been adopted for VAT-related tasks. The technology is now expanding to include more sophisticated applications, further enhancing tax collection systems.
Parliament Discusses AI Integration in Tax Systems
During an October parliamentary session, Italian officials detailed how AI will be integrated into the tax authority’s risk analysis and compliance systems. AI will be used primarily in the pre-assessment stage, where VAT refund claims are first reviewed for inconsistencies. The AI will not influence audit or enforcement decisions but will aid auditors in identifying anomalies more efficiently.
The goal is not to replace human employees but to assist them in handling high-risk or complex cases. AI’s role will be to organize and evaluate large amounts of transactional and taxpayer data, helping auditors pinpoint potential issues.
Machine Learning and AI Handle Vast Amounts of Data
The Italian Revenue Agency’s central tax data warehouse will serve as the primary repository for VAT-related data. This includes:
- E-invoices from the Sistema di Interscambio,
- Cross-border transaction reports,
- Customs import/export declarations, and
- Banking and withholding tax data.
AI and machine learning models are used to process this data quickly. These models clean duplicates, tag counterparties, and align data for more efficient analysis.
VAT Anomaly Detection and Risk Scoring (Pre-assessment Phase)
Once refund requests are submitted, they are scored for risk before being reviewed by human auditors. The algorithms use various techniques, such as:
- Pattern recognition: Identifying spikes in refund claims or suspicious transaction chains.
- Network analysis: Flagging connections to high-risk traders or carousel fraud.
- Historical profiling: Based on past refund history or sector averages.
- Textual AI: Analyzing invoice narratives and tax code usage for unusual patterns.
AI ranks the claims based on their likelihood of error or fraud, feeding dashboards for auditors to make final decisions.
AI Cuts VAT Recovery Processing Times
By 2026, low-risk VAT refunds will be fully automated, cutting processing times and providing businesses with faster liquidity. This will free up time for auditors to focus on more complex cases. The early results show that the average refund time dropped to 67 days by September 2025, ahead of the 70-day target set by the Ministry of Finance.
Italian officials have stressed that their AI approach remains cautious. Unlike other EU countries using automated scraping techniques, Italy’s Revenue Agency avoids using online data to respect privacy standards. Future digital tax frameworks will ensure that privacy protections remain intact.
