DEV Community

TechPulse
TechPulse

Posted on

SEBI to issue AI and ML guidelines for capital markets with kill switches and human oversight

#sebi #ai #regulations #india

SEBI Chairman Tuhin Kanta Pandey told an industry gathering on October 10, 2026, that the regulator will shortly issue guidelines for the responsible use of artificial intelligence and machine learning in India's capital markets. The framework is expected to follow a tiered approach with clear accountability, data controls, kill switches, and human oversight, and to align with the International Organization of Securities Commissions' supervisory toolkit for AI.

What Happened

Pandey spoke at the second Capital Market Confluence 2026, organised by the Bombay Stock Exchange Brokers' Forum. He said the forthcoming AI guidelines would not allow technology to dilute accountability. "Regulated entities remain responsible for AI outcomes, data quality, model governance and cyber resilience," he stated, according to reports from ANI and the Economic Times.

The guidelines are still forthcoming; they have not been published. Pandey also said SEBI expects to issue rules on the settlement price of derivatives on expiry days within about a week, after reviewing feedback on a consultation paper that followed a look at the Closing Auction Session.

Details

The tiered approach is intended to match regulatory intensity to the risk of the AI or ML system. Systems that directly affect clients or trading outcomes, such as algorithmic trading or robo-advisory tools, are expected to face stricter requirements than those used only for internal operations like compliance monitoring or cybersecurity.

Required safeguards include kill-switch mechanisms so systems can be stopped if they behave unexpectedly, and humans-in-the-loop controls so meaningful human oversight remains in place for specified processes. Data controls are also part of the package. The framework will draw on IOSCO's supervisory toolkit to keep governance agile.

These points come from Pandey's remarks as reported by multiple Indian outlets. The full text of the guidelines has not been released, so the precise definitions of tiers, the exact scope of "kill switch," and the obligations on third-party vendors remain to be seen.

Why It Matters

India's capital markets are large and growing. Pandey noted market capitalisation around Rs 472 lakh crore, more than 15 crore unique investors, mutual fund assets near Rs 87 lakh crore, and outstanding corporate bonds around Rs 61 lakh crore. AI tools are already used for surveillance, fraud detection, risk assessment, and investor servicing. Without clear rules, opacity, bias, data issues, and accountability gaps could undermine trust.

SEBI has previously issued a consultation paper on guiding principles for responsible AI/ML use and has reporting requirements for AI/ML systems used by intermediaries. The new guidelines appear intended to move from inventory and principles toward operational controls.

Context

Pandey has spoken about AI rules on earlier occasions in 2026, including at FICCI events, consistently emphasising accountability that cannot be outsourced to technology providers, human oversight, and kill switches. NSE has already required trading members to report AI/ML applications. The latest remarks fit a pattern of progressive tightening as adoption rises.

Globally, securities regulators are grappling with similar issues. IOSCO has produced supervisory materials that SEBI says it will align with. The Indian approach is being described as tiered and risk-based rather than a one-size-fits-all ban or free-for-all.

Impact

For brokers, exchanges, clearing corporations, depositories, mutual funds, and other SEBI-regulated entities, the guidelines will likely require documented model governance, testing, bias checks, data-quality controls, and the ability to halt an AI system quickly. Entities that procure AI tools from vendors will remain responsible for the outputs and for data protection; contracts may need to preserve regulator access.

For investors, clearer rules could reduce the risk that opaque models produce unfair or unstable outcomes in trading, advice, or surveillance. For AI vendors serving the Indian market, the rules may raise the bar on auditability and kill-switch capability.

What Next

Watch for the formal guidelines to be issued by SEBI. Until they appear, the precise obligations, timelines for compliance, and any transitional arrangements are not public. SEBI is also expected to release the expiry-day derivatives settlement guidelines within roughly a week and a circular on simplified digital onboarding for overseas investors after receiving more than 400 comments.

TechPulse Takeaway

Pandey's October 10 remarks are a confirmed signal that SEBI will issue AI/ML guidelines shortly, built around a tiered risk approach, entity-level accountability, data controls, kill switches, and human oversight, and aligned with IOSCO materials. The guidelines themselves have not been published, so claims about exact requirements should stay labeled as forthcoming until the document is out. The statement is consistent with earlier 2026 speeches and with the regulator's existing reporting regime.

Sources

Top comments (0)