In the Income Tax Assessment Procedure, the assessment of officers (AO) is given a full nine months to issue information-seeking notifications to investigate and verify the claims of costs, deductions and income, as explained by taxpayers in their income tax returns.
Recently completed reviews for the tax year 2022-23, however, suggest that a large part of this nine-month period was spent in hibernation, followed by the sudden attack of definitive show-cause forgments that were published close to the Deadline of 31 March.
Random claims
The experience of the taxpayers in the past four years, since the introduction of faceless assessment procedures, has generally been good, with reasonable and honest assessment assignments that have been adopted. But in 2025, the checks and assessment assignments indicate growing cases of gross time abuse. Strange and over -range assessment queries also indicate in the tax authorities that assessing authorities could possibly use generative AI/chatgpt.
In some of the assessment notices, the taxpayers have even been asked to file details of their monthly ration expenses, including flour, rice, spices, oil, gas, electricity, clothing, shoes, polishes, haircuts, cosmetics, perfumes, student fees, car running expenses, rent, insurance, restaurant visit expenses, etc.
The practical inability of the taxpayers to provide such extensive and intrusive details of personal expenses has been used to make random additions to the assessment assignments.
In some of the cases, the AOS valuation experts became and worked such high valuations from those even large trader bankers would give a run for their money.
In one case, the AO replaced the actual repurchase reasons that were paid by the taxpayer company with an imaginary and considerably higher figure-moodschien with the help of a number of AI-generated share valuation method that was completely non-consistent with the applicable legislative provisions and an enormous and random addition. To test the theory. Whether generative AI was used for the valuation method, I asked an AI platform to calculate the valuation for the same case and the exactly the same figures and methodology threw.
In another case, the AO dealt with the long -term profits on the sale of real estate as the normal business income of the taxpayer and the benefit of indexation only denied on the basis of the fact that the aforementioned sale was carried out by an agreement to sell and not with registered deed of sale. The taxpayer had no business history of the trade in real estate.
Another problem that faceless assessments are too much testing is that many cases are transferred to the Jurisdictional AOS on ad hocbase without providing some reasons.
When Jaos issues random orders after such transfers, taxpayers have no choice other than to submit written petitions for high courts or regular appeals for the Commissioner (appeal), which has 600,000 appeal.
Correct training required
The assessment procedures of income tax are quasi-judicial procedures and must be done as such in a fair, just, just and judgmental manner, with the correct application of the mind. AOS must perform a timely, provisional objectivity and materiality analysis to enable them to request specific and relevant material questions from taxpayers in a timely and proactive way. Taxpayers must have enough time to scan and upload their answers and to support documentary evidence.
Where AI tools are used by AOS, they must be used to supplement their natural intelligence and they should not be fully familiar with framing reviews. Furthermore, the generally proclaimed benefit of face-free assessments is the random allocation of cases to non-jurisdictional AOS to abolish the element of Bias and to protect taxpayers against vulnerability.
In order to really work, face -free AOs in smaller cities must be trained correctly to efficiently process complex matters with high -quality transactions.
Views are personal.
Mayank Mohanka is the founder of Taxaaram India and a partner at SM Mohanka & Associates.
