The Austrian Ministry of Finance has announced that the use of artificial intelligence has generated approximately EUR 185 million in tax revenue in 2023 from income tax, corporation tax, and value-added tax.

The Predictive Analytics Competence Centre (PACC), a special unit within the Ministry of Finance, uses predictive analytics and artificial intelligence to detect tax evasion and identify specific entrepreneurs and individuals. In 2023, the PACC’s risk models analysed a total of around 6.5 million cases from a wide range of tax administration sectors. The special unit made discoveries in all sectors of the tax administration. It uncovered false information in employee tax assessments as well as fraud attempts in income tax, corporation tax and value added tax. As a result, the investigators realised around EUR 185 million in tax income. In addition, around 27.5 million cases were investigated with regard to compliance violations, primarily involving the identification of unduly applied for and received subsidies and benefits as well as bogus companies.

Finance Minister Magnus Brunner said, “It goes without saying that we also utilise the possibilities of artificial intelligence in the pursuit of tax evaders and tax fraudsters. The work of the PACC leads to the detection of tax evaders every year and thus contributes to fair competition and compliance with our laws in Austria. I would like to thank my colleagues at the PACC for their valuable work, which, in combination with the work of the auditors, is reflected in this high surplus.”

Of the 34 million cases checked in total, the PACC analyses showed 375,000 cases to be implausible. These were subsequently scrutinised more closely by case handlers and reviewed where necessary.

Christian Weinzinger, Head of the PACC special said, “The PACC was able to use advanced predictive analytics and machine learning methods to help increase additional tax revenue through audit measures and identify individuals and legal entities that were not compliant. The teams at the PACC specialise in various analytical sectors, from predictive and advanced analytics to tax analytics and customs analytics. We therefore cover a relatively broad spectrum of topics.”

Auditing through data analysis and predictive modelling
In the audit sector, various selection methods are used to identify potential cases with a high probability of additional revenue. This includes the selection of cases based on historical audit results, financial data and other relevant information. The same applies to the special VAT audits. Similar approaches are also used for the PLB audit service, which focuses on tax areas such as wage tax and social security contributions, including the review of Covid-19 short-time working. New procedures and techniques such as text mining are used for tax audit case selection to improve the efficiency of audits. Similarly, various models and analyses are used for tax audits in the customs area in order to identify and review potential risks in connection with international trade transactions.

“The PACC has implemented numerous programmes and initiatives to identify and address risks in the tax audit, customs control and other relevant sectors. These measures include both ex-post approaches, which are based on analysing past years, as well as real-time and ex-ante approaches, which aim to identify potential risks in real time or in advance. In addition, various other topics such as digitalisation, cryptocurrencies and international data exchange programmes will be addressed in order to improve the efficiency and effectiveness of tax administration,” said Weinzinger:

AI systems analyse companies right from the start

In the sector of real-time audits, modern machine learning methods are used to carry out risk assessments and optimise the selection of audit cases. For example, business start-ups are also checked in advance to identify entrepreneurs and the people behind them with a higher risk of tax evasion. Modern analysis techniques and network analyses are used to prevent missing-trader fraud, among other things.

The PACC also analyses data from e-commerce platforms and the international exchange of tax information.

“These projects and initiatives show the tax administration’s commitment to an efficient and modern tax administration that is up to the current challenges of the digital economy and global financial markets. Despite all the challenges, the PACC has successfully developed and implemented solutions that further improve the efficiency and effectiveness of customs and tax controls. We use tax income to finance our security, kindergartens, roads and services for families and cultural institutions, among other things. The PACC makes a significant contribution to ensuring that honest taxpayers are not the ones who are stupid and that we put a stop to tax evaders,” said Brunner:

In the future, the PACC plans to further expand its analytical capabilities and initiate new projects. The aim is to further increase the efficiency and effectiveness of the Austrian tax administration and to consolidate the PACC’s role as a central institution for analytical expertise and fraud prevention. Planned initiatives include expanding the use of machine learning methods to include, for example, generative AI using large language models and deepening cooperation with national and international partners. The PACC thus remains a key player in the further development and optimisation of tax administration in Austria.