Introduction
A central challenge in India’s tax policy has been its persistently low tax-to-Gross Domestic Product (GDP) ratio along with widespread tax evasion. Between 2001 and 2022, India’s average tax-GDP ratio stood at 16.36%, the lowest among emerging and developing economies.[1] It is estimated that nearly 4.3% of tax revenues are lost annually due to evasion.[2] At the India AI Impact Summit (February 2026), global leaders and technology experts praised India’s progress in harnessing Artificial Intelligence (AI) to address real-world challenges.[3] One key area where AI’s application deserves attention is tax revenue mobilisation and governance. In this regard, the Income Tax Department’s Project Insight (PI) initiative which is designed to utilise AI and data analytics to strengthen tax administration and improve revenue collection, calls for critical evaluation.[4]
The Advantages
Initiated in 2017 and becoming fully operational in 2019, Project Insight seeks to promote voluntary compliance, minimise high-risk tax evasion cases, ensure fairness and equity in enforcement, and reduce bias in tax administration.[5] The initiative consists of three main components. The Income Tax Transaction Analysis Centre (INTRAC) acts as the analytical core, employing AI and advanced data analytics to process financial data from diverse sources such as banks and financial institutions, property and securities transactions, credit card and GST payments, and high-value purchases, thereby creating a comprehensive financial profile of taxpayers.[6] This enables the Income Tax Department to identify discrepancies between declared income and actual financial activities.
The Compliance Management Centralised Processing Centre focuses on behavioural compliance by using insights generated from INTRAC. It encourages taxpayers who have filed inaccurate returns to correct them through the Non-intrusive Usage of Data to Guide and Enable (NUDGE) strategy, which involves sending SMS or email reminders prompting individuals to align their tax payments with their actual economic activity. Taxpayers may either revise their returns or choose to retain their original filings.
The adoption of AI in tax administration offers several advantages. It helps authorities accurately evaluate taxpayers’ risk profiles and detect potential evasion. It also allows prioritisation of cases based on the scale and complexity of evasion. Additionally, AI can automate routine administrative functions, enabling tax officials to focus on tasks requiring greater human judgment. Further, it improves taxpayer services by assisting with accurate return filing, addressing queries through smart chatbots, and helping prevent tax-related fraud.[7]
PI Initiative: results and concerns
The Project Insight initiative is beginning to deliver visible results. After receiving behavioural nudges, many taxpayers have used the updated return facility to voluntarily correct their original filings. Since the financial year 2020–21, more than one crore revised returns have been submitted, generating additional tax revenue of about Rs.11,000 crore. In a targeted campaign on foreign income and assets, 62 percent of the 19,501 taxpayers contacted revised the information they had initially reported. Additionally, 30,161 taxpayers disclosed overseas assets worth Rs.29,208 crore and foreign income of Rs.1,089 crore, including earnings from cryptocurrencies and other virtual digital assets.[8]
The broader nudge campaign, which reached around 6.25 lakh taxpayers, led to corrections in false deduction claims worth Rs.963 crore related to political donations, along with the payment of Rs.410 crore in additional taxes. Efficiency has also improved, with the average time for processing tax refunds dropping sharply from 93 days to just 17 days.[9] At the same time, the Income Tax Department, using big data analytics and artificial intelligence, identified large scale tax evasion practices in the restaurant sector. Since 2019–20, restaurants were found to have suppressed sales worth Rs.70,000 crore through various manipulative methods.[10]
Similar AI driven systems have already been successfully implemented in countries like Australia, Italy, the United Kingdom, and the United States, where they have helped increase tax revenues.[11] However, as India moves towards algorithm-based tax governance, several important operational, ethical, and legal challenges need careful consideration.
One major concern relates to the quality and origin of data. AI systems are only as reliable as the data they rely on. While they are effective in identifying unusual patterns, they may struggle to distinguish between actual tax evasion and legitimate financial complexity.[12]
Another issue is the risk of algorithmic bias. AI systems trained on past enforcement data can unintentionally replicate existing social or regional biases, leading to disproportionate scrutiny of certain groups or areas. This concern has been highlighted internationally, including in the Dutch childcare benefits controversy.[13]
There are also concerns regarding transparency and due process. Taxpayers must be informed about why they have been flagged, how their data is being used, and how decisions are made. They should also have a clear mechanism to challenge such decisions, along with human oversight.[14]
Data privacy and security form another critical area of concern. The collection and processing of sensitive financial and personal data increase the risk of misuse or cyberattacks.[15]
Absence of Ombudsperson
Finally, India currently lacks an independent AI ombudsperson to review disputed decisions. There is also a need for mechanisms such as algorithmic impact assessments, public disclosure of error rates and appeal outcomes, and regular external audits. Without these safeguards, the system risks undermining trust and fairness in the tax system.[16]
Notes:
[1] Ministry of Finance, Department of Revenue, Tax-GDP Ratio Data.
[2] Central Board of Direct Taxes (CBDT), Direct Tax Statistics Report.
[3] Government of India, India AI Impact Summit 2026 Proceedings.
[4] Income Tax Department, Project Insight Overview.
[5] Ministry of Finance, Annual Report on Direct Taxes.
[6] Income Tax Department, Data Analytics and Compliance Framework.
[7] Government of India, Use of AI in Tax Administration (Official Statements/Reports).
[8] Income Tax Department, Government of India, Official Reports and Press Releases on Project Insight.
[9] CBDT Annual Report and Income Tax Department Statistics on Refund Timelines.
[10] Ministry of Finance, Government of India, Data analytics reports on tax evasion in restaurant sector.
[11] OECD (2021), Tax Administration 3.0 and AI adoption in global tax systems.
[12] Kroll, J. A. et al. (2017), Accountable Algorithms, University of Pennsylvania Law Review.
[13] European Commission Report on Dutch Childcare Benefits Scandal (2020).
[14] OECD Principles on Artificial Intelligence (2019).
[15] Justice K.S. Puttaswamy v. Union of India, (2017) 10 SCC 1.
[16] NITI Aayog (2021), Responsible AI for All – India’s AI Governance Framework.
Author Bio
