Summary
- Issued 20 Jun 2025 for public comment (deadline 11 Jul 2025), based on a SEBI working group's report and inputs from SEBI's Committee on Financial and Regulatory Technologies (CFRT).
- Proposes five guiding principles (para 5): Model Governance, Investor Protection and Disclosure, Testing Framework, Fairness and Bias, Data Privacy and Cyber Security.
- Proposes a tiered approach (para 6): full rules for AI that directly affects clients (algo trading, portfolio management, advisory); a "regulatory lite" subset for internal uses like compliance, surveillance and cyber security.
- Status (as of 2026-10-19): still a consultation paper. No final SEBI guidelines or circular on responsible AI/ML usage appear in SEBI's listings. A secondary source reports SEBI's Chairman said on 10 Oct 2026 the guidelines will be issued shortly (not verified on sebi.gov.in).
Who it applies to
- "Market participants" regulated by SEBI: stock exchanges, clearing corporations, depositories, brokers, mutual funds and other intermediaries using AI/ML.
- Covers both AI built in-house and AI bought from vendors; vendor AI is treated as the participant's own (para 5.1(e)).
- Not extraterritorial. Foreign AI vendors are reached only through contracts with Indian participants.
Key dates
- SEBI circular: AI/ML reporting by market intermediaries
- SEBI circular: AI/ML reporting by Market Infrastructure Institutions
- SEBI circular: AI/ML reporting by Mutual Funds
- SEBI consultation on assigning responsibility for use of AI tools
- Consultation paper on responsible AI/ML usage issued
- Public comment deadline
Full date record
| Date | Event |
|---|---|
| 2019-01-04 | SEBI circular on AI/ML reporting by market intermediaries |
| 2019-01-31 | SEBI circular on AI/ML reporting by MIIs |
| 2019-05-09 | SEBI circular on AI/ML reporting by mutual funds |
| 2024-11-13 | SEBI consultation proposing that regulated persons be solely responsible for AI tools they use |
| 2025-06-20 | This consultation paper issued |
| 2025-07-11 | Comment deadline |
| 2026-10-19 | No final guidelines found on sebi.gov.in |
Penalties and enforcement
- None in this paper; it only proposes principles.
- Separately, SEBI's board considered in Dec 2024 amending the Intermediaries Regulations so that a regulated person using AI tools is solely responsible for their output and data handling (board memo on sebi.gov.in). Secondary sources report this was notified in Feb 2025; the notification itself was not checked here.
- Existing SEBI circulars (2019) already require AI/ML system reporting.
What it means for an Indian company
- Brokers, AMCs, exchanges: start building to the draft now. Key proposed asks:
- Skilled internal team and a designated senior manager overseeing AI (5.1(a), 5.1(d)).
- Fallback plans if AI fails (5.1(c)).
- SLAs with AI vendors; vendor AI counts as your own (5.1(e)).
- Ongoing monitoring and periodic sharing of model accuracy results with SEBI (5.1(f)).
- Independent audit of AI systems, with findings shared with SEBI (5.1(h)).
- Full-verbosity logs so events can be reconstructed (5.1(k)).
- Client disclosure of AI use in client-facing operations, covering features, risks, limitations, accuracy, fees and data quality (5.2).
- Testing in a segregated environment, shadow testing, and keeping model documentation and input/output data for at least 5 years (5.3).
- Fairness checks and bias controls on data (5.4); AI data privacy and cyber policy, and reporting of glitches and breaches (5.5).
- AI vendors to the securities sector: expect SLA performance indicators, audit cooperation and possibly vendor-name reporting to SEBI for concentration monitoring (Annexure B).
- Source quirk: para 6 lists point "5.1(n)" in the regulatory lite subset, but para 5.1 in the paper only runs to (m).
Obligations
Full obligation breakdown coming
We break the most requested policies down into numbered obligations mapped to controls. Leave your details and we’ll tell you when SEBI AI/ML Consultation 2025 is done.
Glossary
- AI/ML: artificial intelligence and machine learning; SEBI defines both in Annexure A.
- MII (Market Infrastructure Institution): stock exchanges, clearing corporations and depositories.
- Shadow testing: running a new model on live data alongside the old process, without letting it make real decisions.
- Regulatory lite: a lighter set of rules for low-impact, internal AI uses.
- Herding: many firms using similar AI models and acting the same way at once, which can amplify market swings.
Key takeaways
- A proposal, not a rule: status is proposed until SEBI issues final guidelines.
- Five principles plus a tiered approach; the heavy obligations hit AI that affects clients.
- Distinctive asks vs RBI FREE-AI: sharing accuracy results and audit findings with SEBI, 5-year retention of model input/output data, shadow testing.
Official sources
- SEBI consultation paper page (20 Jun 2025), accessed 2026-10-19
- Consultation paper PDF, accessed 2026-10-19
- SEBI circular: AI/ML reporting by Mutual Funds (9 May 2019), accessed 2026-10-19
- SEBI board memo on assigning responsibility for AI tools (Dec 2024), accessed 2026-10-19 (listing only checked, not read in full)
This is general information, not legal advice. Last verified 19 Oct 2026 against the official source. Laws change; confirm the current position with qualified counsel before acting.