The Institute of Chartered Accountants of India (ICAI) has partnered with home-grown start-up Sarvam AI to develop a dedicated large language model (LLM) tailored specifically for chartered accountants (CAs). The institute aims to create a domain-specific, sovereign AI platform that enables CAs to leverage AI while safeguarding confidential client information, ICAI president Prasanna Kumar D. said on Friday.
Currently, many CAs use public AI models such as ChatGPT, Gemini and Claude for audit work, financial statement analysis, tax advisory and compliance, raising concerns over data privacy and confidentiality. ICAI’s proposed LLM addresses these risks by providing a secure AI environment where sensitive data remains protected and is not exposed on public platforms. “The idea is to ensure that any work performed by CAs using the AI model remains within the institute’s ecosystem and does not leave the platform,” Kumar said.
Expanding Multi-Tier AI Capacity
Along with the Sarvam AI tie-up, ICAI is also exploring partnerships with OpenAI and Microsoft to strengthen AI training for its members. ICAI officials said that the institute has steadily expanded its AI capacity-building efforts through a structured training programme. While the entry-level programme provides practical training on core AI fundamentals and how to use current tools, the second level enables CAs to build customised AI tools tailored to their individual firms without relying extensively on software developers. The strong response from members prompted the launch of the third level under which CAs will get practical training on working with LLMs – rather than just using AI applications.
“Once we introduced the second level, CAs across practices of all sizes – small, medium and large – began developing AI tools suited for their own firms. That created strong demand for a more advanced programme, which is why we have now launched Level 3 today,” Kumar said.
Data Sovereignty Obstacles
Launching the AI Innovation Summit 2026, Union culture & tourism minister Gajendra Singh Shekhawat said that ICAI has the potential to emerge as the world’s “Institute of Global Trust” by spearheading the development of AI assurance standards. However, he said that India will have to overcome fundamental challenges around data sovereignty, algorithmic bias and trust if it is to fully harness the technology.
“The biggest challenge before India will be data sovereignty,” the minister said, adding that data has become the most valuable strategic resource in the AI era. He warned that algorithmic bias could emerge as another major concern, particularly when AI systems are deployed in insurance, corporate ratings and tax assessments.
“Chartered accountants will no longer be confined to auditing financial statements alone. Their role will expand to independently examine AI systems, validating algorithms, checking for data corruption and safeguarding ethical governance,” the minister said.
Experts said that the latest ICAI initiatives are a step forward in its AI strategy, building on a series of interventions introduced over the past two years. In 2024, the institute launched “CA GPT – Industry Forum”, a platform that integrated the annual reports of nearly 5,000 listed companies, allowing CAs to access and analyse financial data with ease. ICAI is now looking to induct AI into its curriculum to make future professionals AI-ready. So far, the institute has trained over 50,000 members in AI, and developed more than 150 GPT-based tools