CAG taps artificial intelligence, machine learning to improve public auditing quality

India's apex auditor CAG is developing a large language model platform to enhance public auditing quality, speed, and depth, and identify procurement concentration and cartel-risk indicators in real-time.

The country's apex auditor CAG is developing a large language model platform to significantly improve the quality, speed, and analytical depth of public auditing, in addition to working to enable the real-time identification of procurement concentration and cartel-risk indicators.

Comptroller and Auditor General K Sanjay Murthy on Wednesday said the auditor would leverage this integrated data and use tools to detect anomalies, predictive risk scoring models, and network analysis for mapping pan-India procurement.

''Our modernised audit systems will identify repeated bid rotations, suspicious clustering of vendors, abnormal pricing similarities, concentration of awards, and patterns of limited competition across tenders, faster and more efficiently,'' he said at an event in the national capital.

Going forward, Murthy said, the vision is to enable near real-time identification of procurement concentration and cartel-risk indicators. ''Future audit analysis may eventually generate dynamic competition-risk heatmaps across sectors, geographies, contractor ecosystems''.

Speaking as the Chief Guest at the 17th Annual Day of the Competition Commission of India (CCI), he said CAG's focused audits on procurement across major verticals have changed from reporting of individual isolated irregularities to identification of larger systemic patterns.

CAG is deploying Artificial Intelligence (AI) and Machine Learning (ML) tools to analyse data in its entirety.

Every year, Murthy said, various offices of the Comptroller and Auditor General of India generate nearly 20,000–25,000 inspection reports containing lakhs of audit observations relating to government expenditure, infrastructure projects, taxation and procurement, among others.

According to him, the CAG LLM is being developed as a sovereign artificial intelligence platform for public sector auditing and institutional knowledge management.

''These LLM platforms are expected to significantly improve the quality, speed, and analytical depth of public auditing by assisting officers in identifying risk areas, designing audit plans, detecting systemic vulnerabilities, and enabling faster sharing of institutional knowledge and best practices across various stakeholders,'' he noted.

Besides, with improved capabilities, CAG aims to build deeper analytical prowess to assess GST systems, FASTag and logistics data-sets, mining dispatch systems, power exchanges, transport networks and public procurement repositories.