CAG to launch in-house AI assistant for auditors

The Comptroller and Auditor General of India (CAG) plans to introduce an in-house large language model on Monday to support financial analysis and routine audit work. Developed exclusively by the Indian Institute of Technology Madras, LLM 2.0 has been trained on around 65,000 internal inspection reports, alongside CAG’s auditing standards and guidelines.

Auditors will be able to enter simple queries to retrieve information from the underlying reports. The platform is intended to complement CAG’s conventional audit management system by handling numerical calculations, preliminary analysis and pattern identification. Separate streams are likely for compliance, financial and works audits. Auditors will not be permitted to upload material to the platform or the cloud, and the model will serve as an assistant rather than conduct audits itself.

CAG is also shifting toward data-led audit planning for 2026-27, using information across departments rather than focusing solely on individual units. Teams are analysing digital benefit transfers, public works and salary payments for red flags. Its first integrated audits are nearing completion, with reports expected soon. The broader programme covers social sector schemes, ease of doing business and ease of living, multimodal transport and foundational literacy.