Current Affairs Topics Archive
International Relations Economics Polity & Governance Environment & Ecology Science & Technology Internal Security Geography Social Issues Art & Culture Modern History

AI-powered tax governance in India and its challenges

March 19, 2026 5 min read Science & Technology Polity & Governance GS2 (Governance E-Governance) GS3 (Science and Technology)
On This Page ▼

What Happened

  • India's tax administration has increasingly deployed Artificial Intelligence (AI) and machine learning tools to detect anomalies, improve compliance, and reduce human discretion — marking a significant transformation from manual, face-to-face assessment to algorithmic governance.
  • The New Income Tax Act, 2025 — scheduled to come into effect from April 1, 2026 — is designed to enable a more algorithm-driven tax system, with simplified language reducing interpretational ambiguity and enabling rule-based, AI-supported enforcement.
  • AI is being deployed across multiple dimensions: the GSTN (Goods and Services Tax Network) flags suspicious ITC claims, the Annual Information Statement (AIS) cross-references 30+ data sources to surface underreported income, and the faceless assessment scheme eliminates physical contact between taxpayers and assessing officers.
  • Significant challenges remain: a growing backlog of pending appeals (from 4,58,015 in 2020-21 to an estimated 5,16,484 in 2023-24), concerns about algorithmic opacity (taxpayers cannot understand why they were flagged), and risks of over-reliance on AI without adequate human oversight or redress mechanisms.
  • Recommended safeguards include: publishing AI system registers subject to independent audit, creating specialised appeals mechanisms for algorithm-driven decisions, and ensuring compliance with the Digital Personal Data Protection Act (DPDPA), 2023.

Static Topic Bridges

Faceless Assessment Scheme and AI in Direct Tax Administration

The Faceless Assessment Scheme (FAS), launched in 2020 under the Income Tax Act, 1961, is India's most significant recent reform in direct tax administration. Under FAS, taxpayers are assessed without any physical interface with tax officers — cases are randomly assigned to officers in different cities via a National Faceless Assessment Centre, eliminating the scope for corruption, discretionary harassment, and jurisdictional biases. AI underpins several dimensions: case selection for scrutiny, identification of high-risk profiles, and cross-verification of reported income against third-party data.

  • Faceless assessment applies to all income tax scrutiny cases; faceless appeals and penalty schemes followed in 2021.
  • CBDT (Central Board of Direct Taxes) — the apex body for direct tax administration under the Finance Ministry — oversees the scheme.
  • AI tools used: anomaly detection on AIS data, TDS mismatch alerts, high-value transaction flags from the Statement of Financial Transactions (SFT) submitted by banks, mutual funds, property registrars.
  • Annual Information Statement (AIS): introduced in 2021; aggregates 30+ data sources including salary, dividends, securities transactions, bank interest, foreign remittances, property purchases.
  • India's taxpayer base has expanded significantly: the number of ITRs filed crossed 8 crore in AY 2024-25.

Connection to this news: The very success of faceless assessment in reducing corruption creates a new challenge — when errors are made by algorithms or data sources, there is no human officer to approach; the opacity of AI decisions creates a governance accountability gap that the article highlights.

GST Network (GSTN) and AI-Enabled Indirect Tax Compliance

The Goods and Services Tax Network (GSTN) is the IT backbone of India's GST system, handling returns filing, tax payment, and compliance verification for over 1.5 crore registered taxpayers. From its inception, GSTN has used data analytics and rule-based checks; increasingly, machine learning models are deployed to detect fraudulent Input Tax Credit (ITC) claims — a major source of revenue leakage estimated at tens of thousands of crores annually.

  • AI-enabled risk profiling under GSTN: suspicious taxpayers with mismatched GSTR-1 (outward supply) and GSTR-2B (auto-drafted ITC) data are flagged for detailed scrutiny.
  • E-invoicing mandate (B2B transactions above ₹5 crore threshold): generates real-time invoice data used for AI cross-verification.
  • Invoice Management System (IMS): newly mandated system that allows recipients to accept, reject, or keep pending ITC claims — improving data integrity.
  • GSTN handles approximately 75-80 crore invoices per month.
  • Fake ITC fraud: DGGI and GST intelligence units have busted cases worth thousands of crores using data analytics.

Connection to this news: GSTN's AI-driven ITC verification is a practical example of how algorithmic governance improves tax compliance — but also one where errors (genuine MSMEs misclassified as fraudulent) can have immediate business-disrupting consequences.

Data Governance, Privacy, and the DPDPA Challenge

The Digital Personal Data Protection Act (DPDPA), 2023 creates a framework for how personal data — including financial data — can be collected, processed, and used in India. AI-based tax governance systems process enormous volumes of sensitive personal financial data, making compliance with DPDPA a governance imperative. The tension between the Revenue Department's need for data for enforcement and a taxpayer's right to data privacy and informed consent is one of the core challenges highlighted in the article.

  • DPDPA, 2023: establishes rights of data principals (individuals), obligations of data fiduciaries (entities processing data), and penalties for breaches.
  • Tax authorities may qualify as "significant data fiduciaries" under DPDPA, requiring heightened accountability.
  • AI system transparency: under principles of explainability, AI decisions (like flagging a taxpayer for scrutiny) should ideally be explainable to the person affected — current systems often are not.
  • Recommended by experts: AI system registers (a public log of what AI systems are deployed for what purpose), independent audits, and redress mechanisms specifically designed for algorithmic decisions.
  • The growing backlog of tax appeals (5.16 lakh cases as of 2023-24) signals that AI-driven case selection may be generating more disputes than the system can resolve — suggesting the need for better calibration.

Connection to this news: The article calls for governance frameworks that keep pace with AI deployment — including DPDPA-compliant data practices, redress mechanisms, and transparency — reflecting a global concern about algorithmic governance applied to a high-stakes domain like taxation.

Key Facts & Data

  • New Income Tax Act, 2025: effective from April 1, 2026; designed to enable AI-driven, rule-based tax administration.
  • Faceless Assessment Scheme: launched 2020; eliminates physical interface in scrutiny cases.
  • Pending tax appeals: 4,58,015 in 2020-21 → estimated 5,16,484 in 2023-24 (growing backlog despite AI-assisted processing).
  • New appeals filed in 2023-24: 1,44,064; disposed: 1,11,506 — net addition of ~32,558 cases.
  • AIS (Annual Information Statement): aggregates data from 30+ sources including salary, dividends, property transactions, bank interest.
  • ITRs filed in AY 2024-25: over 8 crore (a record).
  • GSTN registered taxpayers: over 1.5 crore businesses.
  • GSTN monthly invoice volume: approximately 75-80 crore invoices.
  • DPDPA, 2023: India's data protection law — directly relevant to AI-based tax governance.
  • CBDT: Central Board of Direct Taxes — nodal body for income tax administration.
  • Key safeguards recommended: AI system registers, independent audits, DPDPA-compliant data governance, specialised algorithmic dispute redress.