OECD: AI Is Cutting VAT Non-Compliance Investigations from 9 Months to 3

AI is dramatically reducing the time tax authorities need to identify and investigate VAT non-compliance, with some administrations cutting investigations from around nine months to just three, according to the OECD.

Speaking at the OECD Tax and Development Days 2026 conference, Hanna Lee, Tax Policy Advisor at the OECD, said AI is helping tax authorities analyse vast volumes of tax data while reducing the time and cost of reviewing information.

Rather than relying on traditional rule-based risk engines, tax authorities are increasingly using AI to detect unusual trading patterns, identify suspicious VAT refund claims, prioritise audit cases and uncover links between businesses and transactions.

The technology is also helping reduce false positives, allowing compliance teams to focus resources on the highest-risk cases.

The announcement provides a clearer picture of how tax authorities are beginning to deploy AI in operational compliance programmes, rather than simply testing the technology.

The development also highlights the growing value of mandatory e-invoicing and digital reporting programmes. As governments gain access to larger volumes of structured transaction-level data, AI is enabling them to analyse information across entire supply chains rather than reviewing individual VAT returns in isolation.

The trend is expected to accelerate as more countries introduce continuous transaction controls and real-time reporting, including under the EU's ViDA reforms.

This content is intended to share insights and practical considerations based on industry experience. It does not constitute legal, regulatory, or financial advice. Regulatory requirements vary by jurisdiction and circumstance, so any compliance-related matters should be reviewed and validated with your own professional advisors.