Indian companies face costly tax and legal errors after generative AI produces fake citations
Indian companies and tax authorities are encountering costly errors after generative AI produced non-existent judicial citations used in tax and legal research. Tribunals and courts, including the Mumbai ITAT and the Delhi High Court, have flagged AI-generated fake rulings, leading to at least one Rs 22 crore tax demand being quashed. CFOs and Big Four advisors say AI is useful for administrative tasks but not yet reliable for complex tax and regulatory judgment without human validation. Firms are running shadow tests, restricting AI to low-risk processes, and emphasizing Responsible AI policies and oversight.
Key Points
- 1Core technical detail: Generative AI hallucinations produced fabricated judicial citations that were incorporated into tax orders and petitions, undermining legal validity.
- 2Business implication: Errors have caused material cost and operational impacts (e.g., a Rs 22 crore demand quashed) and eroded trust, driving firms to limit AI to administrative workflows and require human review.
- 3Future impact: Expect wider adoption of shadow testing, stricter governance and Responsible AI policies, and delayed full automation of tax/legal functions until models, tooling, and auditability improve.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
