Preparing India’s account aggregator ecosystem for the age of AI: Sahamati Labs releases governance framework
As artificial intelligence becomes deeply embedded in financial services, Sahamati Labs, the research and innovation initiative of Sahamati, the Self Regulatory Organisation (SRO) for the Account Aggregator ecosystem, today released A Framework for AI Agents in the Account Aggregator Ecosystem, a design paper proposing how AI agents processing consumer financial data can be governed with the same principles of trust, accountability and consumer protection that underpin India’s Account Aggregator framework.
The paper addresses a critical question for the future of digital finance: How can regulated ecosystems ensure accountability when AI agents, rather than humans, begin making decisions on consented financial data?
India’s Account Aggregator framework established a trusted foundation for consent based financial data sharing by ensuring every data exchange is authorised, traceable and auditable. As AI agents increasingly undertake tasks such as underwriting, fraud detection, verification and financial analysis, the challenge shifts from governing data sharing to governing what happens after the data is received. While the existing framework can establish that data was shared lawfully, it cannot verify how AI systems process that data, what decisions they make, what inferences they draw or whether they remain within the scope of the consumer’s consent.
The paper proposes extending the existing trust architecture through processing aware consent, verifiable identities for AI agents, independent trusted execution that generates cryptographically verifiable records of agent activity and standardised safety evaluation before agents operate within the ecosystem. Together, these principles seek to make AI processing observable, attributable and auditable across institutions.
B G Mahesh, CEO, Sahamati, said, “India has demonstrated that digital public infrastructure can create trust at ecosystem scale. As AI becomes integral to financial services, we have an opportunity to build governance into the ecosystem from the outset rather than after the fact. Trust must extend beyond how data is shared to how AI systems use and act on consumer data. We hope this paper serves as a catalyst for an industry wide conversation on responsible AI in financial services.”
Kiran Gopinath, Chief Innovation Officer, Sahamati and Head, Sahamati Labs, said, “The next phase of digital finance will be defined not just by the availability of AI, but by the trust we can place in it. As AI agents begin acting autonomously on behalf of institutions and consumers, they must operate within the same principles that made the Account Aggregator ecosystem successful- explicit consent, verifiable identity, accountability, and auditability. They must be embedded into the architecture from the outset. This framework extends those foundational trust principles into the age of AI, with the goal of ensuring that AI agents are not only intelligent and capable, but also verifiable, accountable, and worthy of consumer trust.”
Sahamati Labs is publishing the paper as a draft for discussion, inviting engagement from regulators, financial institutions, technology providers and the broader open finance and AI governance community. It sets out the foundational principles for governing AI agents while leaving detailed architecture and operational standards to future collaborative work.