AI-generated GST evasion notices stir human versus machine debate
In the previous financial year alone, AI was responsible for identifying GST mismatches to the tune of Rs 30,000 crore, of which the finance ministry has recovered Rs 15,000 crore so far, a senior ministry official told Moneycontrol.
Central tax officers are said to have sent around 33,000 GST notices so far to businesses for discrepancies in GST returns for the period pertaining to 2017-18 and 2018-19. The recent spike in the number of notices has sparked concerns over increased scrutiny of businesses. In fact, another senior government official confirmed that AI is primarily responsible for the increase in detection of tax evasions, leading to a higher number of notices.
While, the Centre has sent 71 show-cause notices to online gaming companies alone, involving GST evasions worth Rs 1.12 lakh crore in 2022-23 and the first seven months of 2023-24, as many as 15 insurance companies, including Bajaj Allianz, Aditya Birla Sun Life Insurance, and HDFC Life Insurance are alleged to have evaded Rs 2,350 crore in GST so far. Other sectors such as aviation and shipping are also under the radar of tax authorities.
Experts say that given the sheer volume of show cause notices from the government pertaining to GST evasions, the authorities should be watchful of relying entirely on this emerging technology, since it cannot substitute the skills of an officer.
One of the expert says that AI should not replace human intelligence, especially when it comes to tax administration.
“Though AI is a great tool to spot trends, unfortunately what is happening now is that the government is blindly using AI to send notices to assessees. Non-application of the human mind before sending a tax notice is also legally problematic as it is considered grounds for squashing a notice,” he told Moneycontrol.
But, according to four officials from the finance ministry’s revenue department, who spoke to Moneycontrol on condition of anonymity, AI is used only to identify mismatches in GST payments, after which the administration takes a call on whether to issue a demand notice for that particular case.
He added, however, claims that a lot of the notices in the case of GST are a result of mismatches flagged by AI alone, with minimal or no application of the human mind.
Another expert, however, acknowledges that use of AI assists in speedy review and correlation of various data points and identification of gaps. But, he said, the government should not “merely rely on the mismatches without inquiring for the reasons with the taxpayers.”
He added that the tax authorities should discuss the relevant factual and legal aspects with taxpayers before issuing notices to make optimum use of AI and emerging technologies.
One way to ensure better communication between the industry and the government at a time when the reliance on AI is increasing, according to an expert, is issuing an intimation notice before a show cause.
“The law does provide an option for departments to issue an intimation notice before issuing a show cause. That is not being used frequently by GST authorities. With increasing adoption of AI, there should be increasing adoption of issuing intimation notices prior to show cause notices, at least for heavier tax demands. That will go a long way to ease friction and ensure that the taxpayer is not taken by surprise.”
The GST law provides for an intimation notice that informs tax payers about discrepancies in their returns and seeks a response on the reason behind the mismatch within 30 days.
“It is a sophisticated stage of tax administration, and it will become more like this. The case law has started falling into place, and legal interpretation is standardised. Data analytics has reduced the number of areas businesses can hide in,” said one of the officials cited above.
