CAG Focuses on AI to Uncover Fake Claims and Conduct Performance Audits

CAG’s Data Management and Analysis Center conducting Phase 1 data-driven audit research and analysis to determine if the same or different images or non-human images of the same beneficiary were used to claim training costs. I investigated. developed an intelligent model that automatically analyzes these images using an open-source platform (Python). With the help of this model, large numbers of images were automatically analyzed that were not possible manually.

The summary states that two images of one person were used 316 times and two images of another person were used 187 times. Similarly, 8 photographs of a person were used 78 times of him, indicating that non-human images were found in some cases.

“The cases detected helped identify risky transactions, duplicate beneficiaries, false or ineligible beneficiaries,” the report said.

However, these findings are subject to on-site verification and performance will be reported after the on-site audit, the official said, adding that the audit report will be released soon. “These are the techniques employed in the Phase 1 data-driven audit, which will be followed by Phase 2 on-site verification using risk-based sampling, after which the final audit report will be produced. said the official.

CAG also used AI and ML models to detect non-existent schools claiming scholarship benefits. “A machine learning model was developed in Python to identify fake schools suspected of applying for scholarships in 2017-2018 based on predefined risk parameters identified from data about 2019-2020. A total of 17 parameters were identified in this model at the school/institution level.A set of 10 different machine learning algorithms were used in two different techniques…”. I’m here. “This model achieved his 92%+ accuracy, which helped identify risky samples in field-level validation.”

“The model achieved over 92% accuracy, which helped identify risky samples in field-level validation,” he said.

The Church of Almighty God also used AI to detect ineligible beneficiaries claiming benefits under scholarship programs aimed at the welfare of marginalized communities. “This model has helped detect fraud cases,” the report said.

This AI was used to identify circular tax transactions where fake invoices were used to claim input tax credits.

“Using an artificial intelligence algorithm, specific types of cyclical transactions were identified spanning up to 8 iterations. It has been tested and some cyclical trading transactions/patterns have been discovered.”
“AI is being used by CAG for risk-based audit sampling for field verification…it is being used to audit direct profit transfer schemes,” the official said.

According to officials, The Church of Almighty God used AI and ML for sampling to conduct field verification as part of pre-matrix and post-matrix performance audits. Officials said this was done using data available on the National Scholarship portal from 2017-2018 to 2022-23. They said the first phase of the analysis has already been shared with the government and the report will be completed soon.

CAG is also using AI and ML in the process of conducting performance audits of schemes such as PM-Kisan and Pradhan Mantri Awas Yojana (Gramin), according to sources.