New Delhi, Oct 15 The Central Government's efforts to build national AI ecosystem will make AI resources accessible to smaller players in financial services beyond large incumbents, a report said on Wednesday.
The report from business consultancy firm Grant Thornton Bharat said that the shift means firms must proactively align their systems with these platforms and contribute datasets as well as models to them.
The IndiaAI Mission, AI Kosh, the DPDP Act, and CERT-In’s cybersecurity mandates are creating the infrastructure of compute, datasets, data protection, and digital rails that financial institutions can now ride on, the report said.
Financial institutions face four structural challenges, such as data quality, infrastructure gaps, talent shortage, and regulatory ambiguity, it added.
For banks and NBFCs, AI becomes a board-level governance issue, demanding model risk management and fairness programmes. The industry must respond by institutionalising model risk management -- building comprehensive inventories of AI systems, developing fairness and explainability programs, and integrating incident reporting into existing supervisory channels, the report noted.
The report highlighted that national initiatives like AI Kosh and the IndiaAI Compute Platform will bridge capacity gaps for smaller institutions. Capital markets should depend on transparency and auditability to build trust in AI-led decision-making. The business consultancy urged insurers and fintechs to innovate under supervision -- with fairness, explainability, and consumer protection at the core.
RBI’s recent report suggests that only 20 per cent of the regulated entities have adopted some form of AI technology and that the technology deployed so far are simple rule-based non-learning AI models and moderately complex ML models, with limited adoption of advanced AI models
"AI is no longer an experimental tool and is now a regulated infrastructure requiring fairness, transparency, and governance," the report mentioned.
Disclaimer: This post has been auto-published from an agency feed without any modifications to the text and has not been reviewed by an editor