The state govt plans to create an AI defence as rising GST input tax credit (ITC) scams cost the state exchequer thousands of crores of rupees. The models based on AI and machine-learning (ML) will help the govt detect fraudulent entities and transactions.
The commercial tax division of the state’s finance department has tied up with BISAG-N (Bhaskaracharya National Institute for Space Applications and Geo-informatics) to develop a comprehensive system to detect ITC scams. Official sources said that with crores of transactions getting registered every year, the use of AI and ML to identify fraudulent ones has become the need of the hour.
A key govt source said that about 29 crore transactions related to input credit get registered in the state every year. “It is almost impossible to manually check and verify the authenticity of every transaction. The govt has decided to seek the services of BISAG to find a workable solution to malpractices in ITC,” sources said.
“BISAG has already developed a large language model (LLM), which is presently under testing. The AI model will require further fine-tuning. All transactions involving ITC will be run through the model, which will prompt the department on ITC transactions and entities that appear to be fraudulent,” sources said.
Sources in the govt said that using AI and ML will help the commercial tax department to crack down more effectively against ITC scams, as use of AI will identify potential fraudulent transactions.
Govt officials added that the BISAG-developed model will also flag irregularities on the part of govt officials. “In some cases, it has been observed that govt employees have been involved in ITC-related scams. This proposed model will flag irregularities on the part of govt officers as well,” said sources. Sources said that the state govt will start using the AI-based system more aggressively once it is fine-tuned and ready for use.
