IRDAI Sets Up a Seven-Member Working Group to Draft India's First AI Governance Framework for Insurance, With a Three-Month Deadline

India's insurance regulator has started work on rules for artificial intelligence in insurance. According to a report in Insurance Business Asia, the Insurance Regulatory and Development Authority of India (IRDAI) announced on June 19, 2026 that it had set up a seven-member working group to examine how AI is used in the sector and to propose a governance framework. The group has three months to deliver recommendations, which the report says are expected to form the basis of the sector's first formal AI governance framework. The recommendations, however, will not automatically become binding regulation.

The group is chaired by Sandeep K. Shukla, director of the International Institute of Information Technology Hyderabad. Its membership combines technology security and insurance experience. It includes serving and former senior officials from CERT-In, India's computer emergency response team, and from Reserve Bank Information Technology (ReBIT), a technology arm of the Reserve Bank of India, along with representatives from a life insurer, a general insurer and a standalone health insurer. Deepak Gaikwad, IRDAI's general manager and chief information security officer, is the member convener.

The mandate is broad. The group is to map current AI adoption among regulated entities, including the systems in use and the existing governance structures around them. It is to propose a framework for ethical, transparent and explainable AI, with a focus on claims processing and fraud detection. Those two areas are where an automated decision can directly affect a policyholder: whether a claim is paid, delayed or flagged as suspicious. The group is also to assess the effect of frontier AI tools on insurance operations, prescribe security controls against AI-driven automated attacks, and examine whether stress tests are needed for AI-related failures.

Further items concern oversight. The group will review how other markets supervise AI in insurance, develop an audit framework covering pre- and post-deployment requirements, and identify skills gaps and capacity-building needs. The report notes that the audit framework could help clarify who is responsible when AI makes an error in claims or fraud decisions, a question for which no framework currently exists in India. That question of accountability recurs in discussions of AI in insurance: if an algorithm wrongly rejects a claim, it is not obvious whether responsibility lies with the insurer, the vendor that built the model or the person who approved its use.

The group follows IRDAI's revised information and cyber security guidelines, issued in April 2026, which require compliance during the current financial year. The report says the working group's recommendations will not automatically become binding regulation, meaning IRDAI would need to take a further step, such as issuing guidelines or regulations, after considering them.

The report adds industry context from GlobalData. A poll of 113 insurance respondents in the first half of 2026 found that close to a quarter believed AI was not yet ready for widespread use in insurance. GlobalData analyst Ben Carey-Evans linked this to limited use cases so far and unresolved liability, saying that "there is concern around who is liable for mistakes made by AI." GlobalData also recorded about 63,293 active insurance AI-related job listings worldwide in 2025, an increase of roughly 50.9% from 2024, which suggests that hiring is growing even as concerns remain.

How does this compare with other regulators? Singapore's regulator has published draft guidelines on AI risk management, proposed in November 2025, and in July 2026 released an industry white paper on runtime safeguards for AI agents. India's approach, a working group that reports within three months, is a first step rather than a final rule. The group's mandate to review other markets suggests it will draw on such examples, but the report does not say which jurisdictions it will study.

Timing is worth noting. A three-month deadline from the June 19 announcement points to recommendations around the end of September 2026. The report available for this article does not say whether the group has delivered, and the sources reviewed contain no later reports, so the status of the recommendations is unknown. Readers should check IRDAI's website for any published outcome or any consultation that follows.

For insurers, the working group is a signal to prepare. Companies using AI for claims triage, underwriting or fraud detection may wish to inventory their systems, document how models are trained and tested, set up human review for adverse decisions, and strengthen security against AI-assisted attacks. For policyholders, the main potential benefit is clarity about how automated decisions are made and who can be held to account. For technology vendors, the audit framework may bring new requirements for documentation and testing.

The limitations of the report should be kept in mind. It describes a mandate, not results. It does not say how often the group meets, whether it will consult the public or what form the final framework would take. Until IRDAI publishes the recommendations and decides what to adopt, the situation remains one of expectations rather than rules.