SEBI is moving toward formal AI rules for India’s capital markets, and the signal is bigger than one regulator’s checklist. Once AI shapes trading, advice, surveillance, and compliance, it stops being back-office software and becomes part of the market itself.
India’s securities regulator is getting ready to put rules around the way artificial intelligence is used in its markets. According to The Economic Times, SEBI chairman Tuhin Kanta Pandey said on Friday, June 12, that the regulator is working on guidelines for responsible AI across the capital-markets ecosystem.
That sounds procedural. It is not. The important part is where SEBI is drawing the line. AI is no longer being treated only as a tool a broker might use to answer customer questions, or a surveillance product sitting inside an exchange. It is beginning to sit inside trading decisions, advisory recommendations, compliance checks, market monitoring, and the infrastructure that moves orders and settlements. At that point, weak governance is not an internal technology problem. It can become a market problem.
The Economic Times report said the planned framework is expected to cover governance, risk management, accountability, and investor protection. Those four words carry a lot of work. They mean someone has to know who approved the model, what data it was trained on, how it behaves under stress, whether it gives different investors different outcomes, and who is responsible when the system causes harm.
India has a practical reason to move early. The National Stock Exchange crossed 13 crore unique registered investors in April 2026, according to a separate Economic Times report, while its client codes stood at 25.7 crore as of April 25. The Times of India reported this month that NSE trading accounts had passed 26 crore. A market with that many retail accounts cannot afford to let automated advice and automated nudges grow first and ask questions later.
SEBI has already been using AI from its own side of the table. In February, The Economic Times reported Pandey saying the regulator was deploying AI tools to track market misconduct in real time, including insider trading, unregistered investment advice, and misleading financial promotions. That matters because the regulator is not looking at AI as a distant technology trend. It is already using the same class of tools that it now wants intermediaries to govern properly.
There is also a scale issue that makes India different from smaller markets. Pandey said last week that equity issuances crossed Rs 4.5 lakh crore in FY26 and corporate bond issuances exceeded Rs 9 lakh crore, according to The Economic Times. More households are entering capital markets, more firms are raising money through them, and more of the system is becoming digital by default. AI will naturally find its way into that plumbing.
The easy version of an AI policy is to ask firms to disclose that they use AI. That will not be enough here. If an advisory app ranks products using a model, if a broker uses AI to flag risky clients, or if an exchange uses machine learning to detect manipulation, the central question is not whether AI was present. The question is whether the firm can explain the decision, test the model, preserve records, and take responsibility for the outcome.
This is where compliance starts to become a business issue. Large brokers, exchanges, depositories, fintech platforms, and AI vendors can build model-risk teams, audit trails, testing processes, and legal review. Smaller firms may find the same requirements heavy. A responsible framework is supposed to protect investors, but it can also strengthen incumbents if the cost of proving compliance rises faster than the cost of building the product.
The U.S. has moved more slowly
The comparison with the United States is useful because the U.S. debate has been louder but less settled. In July 2023, the Securities and Exchange Commission proposed rules aimed at conflicts of interest from predictive data analytics used by broker-dealers and investment advisers. Axios reported at the time that the proposal focused on AI and similar technologies that could push investor behavior in ways that benefit the firm. The SEC also brought AI washing cases in 2024, with The Wall Street Journal reporting that Delphia and Global Predictions agreed to pay a combined $400,000 over allegedly misleading AI claims.
Those actions show the SEC has not ignored AI. But the U.S. argument has often stayed around disclosure, conflicts, and enforcement after a claim goes too far. SEBI’s posture, at least from Pandey’s remarks, looks more like an attempt to set operating expectations across the market before the most serious failures arrive.
That could make SEBI’s framework more important than its first draft suggests. Fast-growing markets in Asia, the Middle East, and Africa face the same basic problem: retail participation is rising, mobile-first investing is normal, and AI vendors are selling tools into every layer of finance. A rulebook that connects governance, accountability, investor protection, and market integrity could travel well, especially in jurisdictions that do not want to wait for the U.S. to settle its own debate.
The risk is that the rules become paperwork. Firms are good at producing policies that say a model is reviewed, monitored, and accountable. Investors need more than that. They need rules that make it clear when a firm cannot hide behind a vendor, when an automated recommendation must be explained, and when a model that works in normal markets has to be restrained during stress.
SEBI has not published the final guidelines yet, so the real test is still ahead. The direction, though, is clear enough. India’s market regulator is treating AI as part of the machinery of finance, not as a software feature sitting politely at the edge of it.
Also read: AI financing is now large enough to move public markets • KPMG pulled an AI report after its own facts fell apart • Anduril wants export controls to catch up with drone warfare
