The GST Council has praised Andhra Pradesh’s use of artificial intelligence and advanced data analytics to identify tax-evasion risks, citing higher detection values, greater transparency and reduced subjectivity in the state’s tax administration.
The recognition came at the 57th meeting of the GST Council in New Delhi, presided over by Union finance minister Nirmala Sitharaman. Andhra Pradesh has introduced AI-enabled systems for scrutiny of returns under Section 61, audit under Section 65 and inspection under Section 67 of the GST Act.
The state’s risk-assessment engine combines GSTN data, e-way bill information and third-party datasets to identify high-risk cases using data-driven parameters. According to the department, the system selected 10,817 cases during its first six months of AI-enabled scrutiny and detected tax evasion worth ₹743 crore. Under the earlier system, the department had detected ₹366 crore in comparable enforcement activity.
Average detection per case also increased from ₹3.08 lakh to ₹27.63 lakh, the department said. The figures indicate a shift from broad or subjective case selection towards concentrating officers’ attention on cases assessed as carrying higher compliance risks.
The system retains human supervision at key stages. Cases can be escalated to the joint commissioner, the regional internal monitoring committee or the state internal monitoring committee. Statutory notices continue to be issued by designated officers, ensuring that AI-supported identification does not replace the formal responsibility of tax officials.
“AI is helping us move from subjective selection to objective, risk-based enforcement,” Babu Ahmad, chief commissioner of state tax, Andhra Pradesh, said. “The technology enables officers to focus their efforts on high-risk cases while ensuring transparency, accountability and human oversight.”
The department has also developed an AI-powered Legal Assistant trained on a knowledge base containing around 22,000 judicial pronouncements and the GST statutory framework. The tool assists officials in preparing legal instructions, paragraph-wise remarks, counter-affidavits and replies, although its outputs require approval from competent officers.
The Legal AI platform currently supports 13,719 cases. These include 8,098 cases before the First Appellate Authority, 577 appeals before the GST Appellate Tribunal and 5,044 matters before the high court. The department said the system had improved consistency and efficiency in litigation without a corresponding increase in legal manpower.
Andhra Pradesh’s model is also drawing interest from other states. Delegations from Rajasthan and Chhattisgarh have visited the state to study its AI-based systems, while Assam, Delhi, Uttarakhand and other states have expressed interest in examining the initiatives. The inter-state interest places the state’s tax technology programme within a wider effort to use digital systems for more targeted public administration while retaining formal oversight by government officials.
