The proposed framework comes as financial institutions increasingly deploy AI-powered tools for investment advisory services, customer onboarding, trading and market surveillance.
Who is responsible when AI advice goes wrong?
One of the biggest concerns is determining accountability when AI-generated recommendations result in unsuitable investments or financial losses.
Rananjay Singh, COO, Enso Group, said regulated financial institutions cannot escape responsibility simply because an investment decision originated from an algorithm.
"SEBI’s proposed AI guidelines arrive at a critical juncture: technology increasingly shapes investment recommendations, onboarding and surveillance, but an opaque algorithm must never become an excuse for an unaccountable decision," Singh said.
He highlighted the scale of potential exposure, noting that NSE reported 13.7 crore unique registered investors in September 2026, covering over 99% of Indian postal codes.
Singh said SEBI had already made regulated intermediaries accountable for AI outputs, including third-party tools, in 2025. The next step should involve enforceable model validation, audit trails, bias testing and escalation protocols.
However, accountability does not automatically mean investors are entitled to compensation for every loss. Responsibility would depend on the applicable regulatory obligations, circumstances of the recommendation and whether misconduct or negligence occurred.
MUST READ: SEBI proposes FPI rule relaxation after ₹5.61 lakh crore outflows: Will foreign investors return?
Can biased algorithms lead to financial losses?
Jitin Thapar, Creator of EPICC Reinvention 'Inside-Out' Philosophy and Global Strategic Advisor, said AI can improve market surveillance, identify abnormal trading behaviour and strengthen risk assessment.
However, these benefits must be accompanied by adequate safeguards.
"In these markets, even the slightest mistake of the algorithm or bias of AI-driven recommendations may lead to certain financial losses for investors," Thapar said.
He stressed that AI-generated recommendations should be explainable, particularly when they influence investment decisions, trading operations and risk management.
Financial institutions should also conduct regular audits to identify algorithmic biases, errors and vulnerabilities while protecting investors' personal information.
ALSO READ: Sebi considering proposals on MTF, says chairman
Why is human oversight important?
SEBI's proposed human-in-the-loop approach seeks to ensure that critical AI-assisted decisions remain subject to human supervision.
Singh warned that financial institutions using similar AI models could inadvertently amplify market volatility if automated systems react simultaneously to incorrect signals.
He recommended stronger safeguards for investor-facing recommendations and trading decisions, while allowing lighter compliance requirements for routine back-office automation.
Thapar also cautioned against excessive regulation that could discourage technological innovation.
"However, regulation must not turn into an obstacle to innovations. It should serve as a basis for the development of such a framework that allows technology to develop responsibly without losing the faith of investors," he said.
For retail investors, the proposed framework could improve transparency and strengthen avenues for questioning unsuitable AI-generated recommendations. However, the extent of these protections will depend on SEBI's final guidelines and their enforcement.
DO READ: SEBI mulls limits on margin trading as retail leverage hits Rs 1.53 lakh crore, says report