What AI disclosure obligations does SEBI's Research Analysts Third Amendment 2024 create?

SEBI (Research Analysts) (Third Amendment) Regulations 2024 (notified December 16, 2024) and its implementing circular SEBI/HO/MIRSD/MIRSD-PoD-1/P/CIR/2025/004 (January 8, 2025) created binding AI governance obligations for all SEBI-registered Research Analysts. The amendment applies to individual and entity RAs — the 1,500+ registered research analysts who publish equity research reports, trading calls, sector analyses, and market commentary for retail and institutional clients in India. Three direct AI obligations: (1) Mandatory AI tool disclosure at terms disclosure: Research Analysts must disclose the extent of use of AI tools in generating research reports and analysis at the point of disclosure of terms and conditions to clients. Unlike the IA amendment's per-advice disclosure requirement, the RA amendment links AI disclosure to the terms and conditions point — when the RA formalizes its engagement with a client, the AI disclosure is part of that formalization. The disclosure must specify the extent of AI use — a disclaimer that says 'AI may be used' without specifying the AI's actual role in the research process does not satisfy the requirement; (2) Full RA accountability for AI-generated research errors: Research Analysts who use AI to generate trading calls, sector analyses, price targets, and market recommendations bear full regulatory accountability for AI-generated errors. An RA cannot disclaim accuracy by attributing an incorrect recommendation to its AI research platform; SEBI enforcement will proceed against the RA, not the AI vendor; (3) Audit trail for AI research including AI inputs and outputs: SEBI's record-keeping requirements for Research Analysts now explicitly extend to AI-generated research artifacts — the AI inputs used (data sources, prompts, parameters) and the AI outputs generated (preliminary research drafts, screening results, analysis outputs) must be maintained as part of the research record.

How does SEBI's RA Amendment 2024 govern AI-generated trading calls in India?

Trading calls — buy, sell, hold, and target price recommendations for specific securities — are the core regulatory product of SEBI-registered Research Analysts. AI-generated trading calls are a rapidly growing practice: algorithmic screening and scoring models generate preliminary buy/sell signals that RAs then publish, sometimes with minimal human review. SEBI's 2024 amendment creates governance requirements for AI-generated trading calls that were absent from the original 2013 RA Regulations: (1) AI trading call disclosure: when an AI system generates a trading call — or generates the primary research supporting a trading call — the RA's publication must disclose the AI's role. A trading call that was entirely AI-generated but presented as human analysis fails the disclosure requirement; (2) AI trading call accuracy tracking: SEBI RA Regulations require tracking of research recommendation performance. When AI generates trading calls, the accuracy tracking must be attributed to the AI-generated recommendation — not averaged into the RA's overall track record in a way that obscures AI-specific accuracy. An AI that generates systematically poor trading calls for small-cap stocks while outperforming on large-caps must be monitored at that granularity; (3) No AI platform defense for insider trading surveillance: SEBI's market surveillance infrastructure (including SAT cases) monitors RA trading calls for potential market manipulation and insider trading. An RA whose AI platform generates concentrated buy calls for securities in which the RA or its affiliates hold positions cannot argue the AI generated the calls autonomously — the RA remains fully accountable for AI-generated market manipulation; (4) Retail investor AI trading call protection: a significant proportion of Indian retail investors rely on RA trading calls published on social media and advisory platforms. AI-generated trading calls distributed at scale — the AI publishes thousands of personalized recommendations to individual retail clients — create a systemic risk that concentrated AI errors could affect many retail investors simultaneously. SEBI's amendment creates an audit trail obligation specifically to enable post-incident investigation of AI trading call errors.

What audit trail obligations do SEBI's 2024 RA regulations create for AI-generated research?

SEBI's RA Third Amendment 2024, as implemented by the January 2025 circular, extends the RA record-keeping obligation to AI research artifacts — creating a specific audit trail requirement for AI inputs and outputs used in research generation. The audit trail requirement has four operational dimensions: (1) AI input documentation: the data sources, parameters, and prompts used to generate AI research outputs must be documented per research artifact. For an AI large language model used to generate a sector analysis, this means logging the specific data inputs (earnings transcripts, market data, management guidance) and the generation parameters used — not merely retaining the final published report; (2) AI output retention: the raw AI output — prior to human review, editing, or approval — must be retained alongside the final published research. This creates an audit trail that enables SEBI to compare what the AI generated with what was published, identifying where human edits introduced material changes and whether those changes were appropriate; (3) AI tool version documentation: when an RA's research platform is updated — AI model version, screening logic, scoring methodology — the version used for each research report must be documented. Research published using an old AI version and research published using a new version may have materially different analytical bases, and SEBI's audit trail requirement must reflect this version-level specificity; (4) AI research chain of custody: for research that passes through multiple AI tools (a screening AI that generates a universe of stocks, a valuation AI that generates price targets, an NLP AI that synthesizes the write-up), the full chain of AI tools used must be documented — creating a research provenance record that SEBI can trace from final publication back through each AI stage.

How does client segregation under SEBI RA Regulations interact with AI personalization of research distribution?

SEBI RA Regulations require Research Analysts to maintain segregation between different client categories — institutional clients, retail investors, individual advisers, and the general public — with different disclosure and content standards for each category. AI-powered research distribution platforms that personalize research delivery create a compliance tension with the segregation requirement: (1) AI personalization vs. content segregation: an AI that personalizes the framing, emphasis, or presentation of a research report for different clients — highlighting growth factors for a retail investor and risk factors for an institutional investor — may create a situation where the same research is effectively different content for different clients, without the formal segregation process required by SEBI RA Regulations; (2) AI distribution timing differential: an AI distribution system that prioritizes delivery of trading calls to institutional clients before retail clients — based on AI-scored engagement or revenue potential — creates an information asymmetry that conflicts with SEBI's fair access requirements. AI personalization logic that systematically delays retail access to material research creates a regulatory issue regardless of whether it is intentional; (3) AI-generated research disclaimers: SEBI requires research reports to include specific disclaimers covering analyst holdings, issuer relationships, and compliance attestations. AI-generated research must include human-reviewed and approved disclaimers — an AI that generates disclaimers automatically without human verification of the specific accuracy of the disclosure (e.g., whether the analyst actually holds the security being recommended) creates a compliance gap; (4) AI large-scale retail distribution: an RA that uses AI to distribute personalized trading calls to hundreds of thousands of retail clients simultaneously — at a scale that was previously impractical — must ensure that the AI distribution system maintains per-client segregation compliance for every distribution event, not only on a sample or average basis.

AI Governance · SEBI · Research Analysts · AI-Generated Research · Trading Calls · India BFSI

SEBI Research Analysts Third Amendment 2024 AI Governance: Mandatory AI Research Tool Disclosure, AI-Generated Trading Call Accountability, Audit Trail for AI Research Inputs and Outputs and Client Segregation with AI Personalization India

SEBI (Research Analysts) (Third Amendment) Regulations 2024 (December 16, 2024) and implementing Circular SEBI/HO/MIRSD/MIRSD-PoD-1/P/CIR/2025/004 (January 8, 2025) mandate disclosure of AI tool usage in research, full RA accountability for AI-generated trading calls, and audit trail obligations for AI research inputs and outputs. Applies to 1,500+ SEBI-registered Research Analysts and AI-powered investment research platforms. Pairs with the SEBI IA Amendment 2024: advisers disclose advice AI, analysts disclose research AI. Zero global competitor coverage.

Zytra Research Team · December 2024 · 11 min read

📖 Part of the India BFSI AI Governance Complete Guide 2026 — RBI, SEBI, IRDAI, NPCI & DPDP in one place.

India's investment research market — NSE-listed equity research from brokerages, independent research boutiques, and the growing AI-powered research platform segment (Tickertape, Smallcase Research, Sensibull, and dozens of SEBI-registered RA entities) — uses AI extensively for stock screening, sector analysis, price target generation, and trading call distribution. SEBI's (Research Analysts) (Third Amendment) Regulations 2024 created the first binding governance obligation specifically for AI-generated financial research: mandatory disclosure, full RA accountability, and audit trail requirements for AI inputs and outputs.

This amendment pairs with the SEBI IA Second Amendment 2024, creating a consistent AI accountability framework across the two primary client-facing advice and research categories: investment advisers must disclose AI in advice; research analysts must disclose AI in research. Both carry sole accountability with no vendor safe harbour.

The Audit Trail Obligation: Research Provenance for AI-Generated Content

SEBI's record-keeping requirement for Research Analysts has historically focused on the final published research report and the analyst's holdings disclosure. The 2024 amendment extends this to AI artifacts — the AI inputs used and the raw AI outputs generated, distinct from the final published form. This creates a research provenance chain that enables SEBI to reconstruct how an AI-generated research recommendation was produced — and whether the AI platform's logic was appropriate, consistent, and non-manipulative.

SEBI RA AI Governance Checklist

☐ Terms disclosure AI clause: AI tool usage extent disclosed at point of terms and conditions; specifies AI's actual role (screening, analysis, report generation, distribution); updated when AI role changes materially
☐ AI trading call accountability: AI-generated trading calls identified and attributed; AI accuracy tracked separately from human-generated calls; no AI platform defense for SEBI enforcement
☐ AI research audit trail: AI inputs (data sources, prompts, parameters) logged per research artifact; raw AI output retained alongside final published form; AI tool version documented per publication
☐ AI research chain of custody: multi-AI research workflows (screening + valuation + NLP) documented end-to-end per research output
☐ Disclaimer AI verification: AI-generated disclaimers reviewed by human for accuracy (holdings, issuer relationships, compliance attestations) before publication
☐ Client segregation AI compliance: AI distribution logic tested for institutional/retail content and timing segregation compliance; no AI-driven information asymmetry
☐ AI retail distribution scale: per-client segregation compliance verified for AI mass distribution events, not only on aggregate
☐ SEBI examination readiness: terms disclosure AI clause, AI trading call records, AI research audit trail, AI tool version log — packaged for SEBI RA inspection