Artificial Intelligence in Tax Administrations

The purpose of this commentary is to review some of the use cases of artificial intelligence (AI) in Tax Administrations (TA).[1]

As a truly relevant report “Tax Administration 2022” published on June 23 by the OECD Forum on Tax Administration (FTA) emerges[2], and as TAs become more comfortable with managing large data sets and computing power increases, the use of AI and machine learning is opening up new approaches in risk management.

More than 70% of the TAs surveyed report that they are already using innovative techniques to exploit data in a way that can uncover previously hidden assets or identify new risks[3].

TAs are making significant investment in digital identity programs, including the use of AI to improve efficiency and effectiveness, which helps to consolidate digital identity as the cornerstone of a successful digitalization activity.

For example, in Sweden, the TAs have been using AI in the registration of companies since May 2021. This AI-based service classifies requests based on a set of established risk factors, and then the requests are processed in different ways, depending on the assigned category.

Previously, the categorization process was manual and time-consuming. After less than a year in operation, the first version of the risk assessment service has already generated a solid return on investment with the business registration process shortened by up to six days and a cost reduction of SEK 28 million (approximately 16% of the total cost of the business registration process).

In France, to improve the detection process of undeclared constructions or developments, the French tax administration (DGFiP) uses AI and data enhancement based on aerial photographs taken by the Institut national de l’information géographique et forestière (IGN)[4].

The algorithms will make it possible to extract the outlines of buildings and swimming pools from the public aerial images of the IGN, which everyone can consult on the website .

A computer process then verifies, from the declarations made by the owners to the tax authorities, whether the items thus detected in the images are correctly taxed by direct local taxes (real estate tax in particular). Next, an agent of the tax authorities systematically checks each detected anomaly before any operation of transfer of the real estate to the owner and then to taxation.

The use of advanced techniques in artificial intelligence, machine learning and machine-to-machine linkages is opening up new service options for TAs that allow more “compliance by design” style approaches to be available, which is a growing trend that is expected to accelerate.

AI is also being used in techniques to predict the behaviour of taxpayers (known as behavioural insights in English).

In Brazil, the Behavioral Insights Project began with a test on 2489 small companies that earned more during the pandemic than in previous years to drive compliance behavior. In this test, there were 3 different letters, each using a different behavioral science technique and a control of the existing card.

The learning was that the inclusion of messages using behavioral techniques in the letters sent to taxpayers makes a difference, depending on the outcome of the taxpayer profiles. As a result, there is now an AI study to predict the right letter choice for each taxpayer.

AI is also expanding its use in the areas of taxpayer services, as emerges from Tax Administration 2022 where 40% of the TAs consulted have a virtual assistant that use AI in some way to improve the service.

In a publication where the topic is discussed in detail, the existence of virtual assistants in the TAs of Spain, Peru, Australia, Canada, the United Kingdom, Ireland, Finland, Sweden, Latvia, Estonia, the Republic of China, Russia, Singapore, Guatemala, Chile, Mexico, Costa Rica, Colombia and Brazil has been highlighted.[5]

As we see and as it happens in private sector companies, AI serves to provide increasingly personalized services to taxpayers.

For example, the Internal Revenue Authority of Singapore (IRAS) has leveraged end-user automation, data, and AI tools to provide seamless, personalized taxpayer services.

IRAS added a digital adoption tool to add personalized step-by-step on-screen guidance on digital services in my Tax Portal.

Taking an agile approach to acting on comments as they arise, IRAS iteratively improved the guide’s content and navigation for the taxpayers.

During the pilot period, IRAS achieved a 13% reduction in contacts requesting guidance on electronic filing and more than 86% of the taxpayers surveyed found it useful in helping them complete their digital transactions.

TAs are also exploring how AI can be integrated into compliance processes across the organization, and this is likely to be critical to the digital transformation of compliance management and risk management in the future.

In Canada, the CRA is empowering its assurance and advisory professionals to incorporate artificial intelligence and data analytics into their work. The CRA has evolved from the use of traditional software, marketed specifically for internal auditors, and has invested in learning to use tools that are more versatile to enhance their analysis. This has given the CRA the opportunity to use the many robust natural language processing, machine learning, and data visualization packages available in these tools to help transform the way the CRA accesses, processes, and analyzes data.

The Tax Authority of Singapore (IRAS) is expanding the use of data and AI across all functions, including through the development of new AI solutions to better manage tax non-compliance and improve service delivery.

These AI initiatives include service quality monitoring as natural language processing models have been developed to qualify 100% of live chats for service quality monitoring.

The new AI solution can increase the objectivity and productivity of service quality monitoring, allowing IRAS to identify chats that require attention and take initiative-taking measures to improve the service experience.

Finally, AI is also being used in human resources, where we can mention that the Tax Authority of Singapore (IRAS) which, for recruitment, uses conversational chatbots with AI to manage and improve the experiences of applicants, for example, 24/7 access to complete the interview process and upload their resumes in a single session.

IRAS designed the conversation flow and questions by leveraging AI, and chatbots can also answer common recruitment queries. IRAS customer satisfaction scores averaged 94% across all job roles.

Likewise, also in Singapore, one tool seeks to match resumes of potential applicants to job postings based on skill requirements, while another tool under development seeks to pilot a more competency-based assessment of job applicants.

In short, the potential of AI in TAAS is enormous, which is why it is extremely useful for TAs that are in the process of adopting technologies to carefully analyze the cases and best practices of others that have already started the path.

AI presents a scenario of disruptive changes for TAs between the medium and long term. Digital transformation involves not only structural but also cultural changes.

TAs need to understand how technology impacts their functions and develop the skills needed to use it efficiently.

But we must never lose sight of the fact that ICTs are a tool to obtain better results, i.e., they are not an objective in themselves; technology is of no use “as a fashion”; we must always ask ourselves what the strategic objective of its incorporation will be.

This whole process of digitalization of TAs, including the adoption of new technologies such as AI, should not be conducted in isolation, but should be integrated into the digitalization of countries, within the concept of digital government.

Governments must collaborate with the different actors involved to ensure the proper use of AI, in an ethical and equitable way, protecting the fundamental rights of citizens and always seeking to ensure that ICTs are an integrating element with the human resources of TAs.

AI should not focus on replacing public competencies but should increase or complement human capabilities so that people can add value to their tasks while improving the quality and efficiency of public functions for citizens.

I am convinced that technology such as AI, by providing transparency and efficiency, can be vital to improve citizens’ trust in governments, especially in times of crisis such as the current ones.

[1] Without prejudice to all those cases of use of AI stated in the CIAT/AEAT /IEF,   Tax Administration Review No. 48 in the article “Artificial intelligence in tax administrations” COLLOSA, ALFREDO (spanish)
[2]  OECD (2022) Tax Administration 2022: Comparative Information on OECD and other Advanced and Emerging Economies, OECD Publishing, Paris,
[3] OCDE et al (2022), Inventory of Tax Technology Initiatives,, Table DM5
[5] LA DIGITALIZACIÓN EN LAS ADMINISTRACIONES TRIBUTARIAS: Con particular referencia a los asistentes virtuales. Estado de la cuestión, en especial, en Iberoamérica” de los autores Domingo Carbajo Vasco y Alfredo Collosa, Editorial Tirant lo Blanch 2022.

Disclaimer. Readers are informed that the views, thoughts, and opinions expressed in the text belong solely to the author, and not necessarily to the author's employer, organization, committee or other group the author might be associated with, nor to the Executive Secretariat of CIAT. The author is also responsible for the precision and accuracy of data and sources.

Leave a Reply

Your email address will not be published.

CIAT Subscriptions

Browse through the site without restrictions. Consult and download the contents.

Subscribe to our electronic newsletters:

  • Blog
  • Academic offer (Only in spanish)
  • Newsletter
  • Publications
  • News alert

Activate subscription

CIAT Members

Representatives, Correspondent and Authorized staff (TA)