AI in Government

AI is helping tax administrations around the world modernise their operations, detect fraud, improve compliance and deliver better services to taxpayers. By analysing large volumes of structured and unstructured data, AI supports smarter audits, faster case handling and more personalised taxpayer interactions. Tax authorities are among the most experienced government actors in applying AI, especially for classic rules-based approaches — but new capabilities bring new challenges, particularly around trust, transparency and the use of sensitive data.

The current state of play

Tax administrations have adopted AI across multiple areas of their operations.

^{Source: OECD Data Explorer – Inventory of Tax Technology Initiatives 2024 (https://oe.cd/dx/ITTI2024).}

Some particular areas of importance involve:

• Improving compliance and detection of evasion and fraud. AI is used to detect complex patterns in tax data, link related entities or behaviours, and analyse unconventional sources such as satellite imagery or even social media. Some administrations now apply AI to detect large-scale refund fraud or undeclared properties and assets.
• Assisting administrative decision-making processes. AI helps streamline internal workflows — for example, by grouping similar cases or distributing them more efficiently. These systems allow routine issues to be automated while reserving expert attention for complex disputes or appeals.
• Enhancing risk assessment processes. Tax administrations use machine learning to score taxpayer behaviour, detect anomalies and prioritise cases for further review. As models become more sophisticated, they help prevent errors before they occur and reduce the need for resource-intensive audits.
• Improving taxpayer services. AI-powered virtual assistants now handle a growing share of taxpayer interactions — from answering basic queries to enabling secure self-service features. Newer tools incorporate large language models to improve accuracy and personalisation.
• Assisting taxpayers in filing. AI is increasingly used to pre-fill tax returns and detect errors at the point of submission. This reduces administrative burdens, improves accuracy and builds confidence in the system.

Despite their progress, tax administrations face challenges around workforce skills, regulatory frameworks and the high costs of developing and scaling robust AI tools. Safeguards are essential to preserve taxpayer rights, ensure explainability and avoid skewed outcomes — particularly given the central role of trust in tax compliance.

Examples from practice

• Greece: Real-time tax enforcement. The Independent Authority for Public Revenue is using AI to detect tax evasion, automate procedures and even identify undeclared swimming pools through satellite imagery — helping auditors act quickly and efficiently.
• France: Detecting undeclared property development. The French tax authority uses AI to analyse aerial imagery and compare it with taxpayer declarations. Suspected anomalies are flagged for human review, improving oversight while ensuring fairness.
• Austria: Predictive analytics for compliance. Austria’s Predictive Analytics Competence Centre analyses millions of tax cases each year. Its models help flag non-compliance, identify errors and guide both retrospective and real-time audits.
• Singapore: Personalised taxpayer services. Singapore’s IRAS chatbot uses large language models and automation to deliver seamless, humanised taxpayer assistance across platforms — cutting wait times and saving over 11,000 taxpayer hours in a single year.
• Brazil: Accelerating tax appeals. Brazil’s AI Litigation Project uses supervised learning and clustering to group similar tax appeal cases and support officers with text suggestions. This helps reduce decision times and tackle large backlogs.
• Brazil: Analysing the outcomes of standard tax letter requests to taxpayer. Brazil’s FAPE programme used AI and behavioural insights to tailor tax letters based on taxpayer profiles and responses. By adjusting tone and messaging, it improves compliance and identifies the most effective communication strategies.

Untapped potential and the way forward

AI has the potential to transform tax administration, not just optimise it. When integrated into digital tax ecosystems — as envisioned in the OECD’s Tax Administration 3.0 framework — it can make processes more seamless and adaptive.

Delivering this requires aligning technical capabilities with legal frameworks, organisational change and public expectations. A focus on rules-as-code, transparency and lifecycle governance can help. The OECD Forum on Tax Administration is piloting a supporting framework, with results expected in 2026.

Learn more

Review a detailed section on AI in tax administration here.