Incident sheet · /enterprise-ai

Filed
Read
4 min · 734 words
Severity
Routine

SEBI to Mandate AI Disclosure and Human Oversight in Capital Markets

SEBI Chairman Tuhin Kanta Pandey says guidelines will force AI disclosure, vendor accountability and human oversight — but no timeline was given.

SEBI will make market firms disclose AI use and keep humans in the loop, chairman says
FeedSEBI will make market firms disclose AI use and keep humans in the loop, chairman says — Pexels / pixabay

Event log

  1. SEBI will issue AI guidelines requiring disclosure, accountability and human-in-the-loop oversight; no timeline announced (Nov announcement, New Delhi).
  2. December 2024 amendment already makes regulated entities solely liable for AI tool outputs and investor data handled by them.
  3. June 2025 consultation proposed 5-year data retention, independent audits and plain-language client disclosures.
  4. SEBI's SUDARSAN tool flagged 20,000+ instances of fraudulent social media content since November 2025.
  5. Pandey expects quantum-computing risks to markets between 2032 and 2035.

India's markets regulator will force capital-market firms to disclose their AI use and keep humans in the loop on critical decisions. SEBI Chairman Tuhin Kanta Pandey announced the plan at the India International Trade Fair in New Delhi on Saturday, PTI reported. The report specifies no timeline. That omission matters.

Pandey said the coming guidelines will set guardrails covering three areas: disclosures, accountability of regulated entities, and a human-in-the-loop approach with human oversight over critical decisions. He said SEBI is reviewing its regulatory framework to ensure AI use does not compromise investor protection, market safety or integrity.

What already binds firms today?

SEBI does not start from zero. In December 2024 it amended regulations to make regulated entities solely responsible for outputs of AI tools they use — whether developed in-house or procured — and for privacy and security of investor data those tools handle. The vendor is not the shield. The regulated firm carries the liability.

The new guidelines build on a consultation paper SEBI released on June 20, 2025, with comments open until July 11, 2025. That paper drew on a SEBI working group calling for a "human-in-the-loop throughout the development and deployment lifecycle."

What did the 2025 consultation propose?

  • Disclosure to investors: Firms using AI in client-facing operations — algorithmic trading, portfolio management, advisory services — must inform clients about the model's features, purpose, risks, limitations, accuracy, fees and data quality, in plain language.
  • Vendor accountability: Third-party AI services "are deemed to be provided by the market participants," who remain legally responsible.
  • Senior oversight: Designated senior management must oversee models, with fallback plans, independent audit teams, and audit findings plus accuracy results shared with SEBI.
  • Testing: Models must be tested in segregated environments, shadow-tested on live traffic before deployment, monitored after launch, with input and output data retained at least five years.
  • Fairness: Models "should not favour or discriminate one group of clients/customers over another."
  • Data privacy: Firms must comply with data protection laws and report breaches and glitches to SEBI.

Internal uses — compliance, surveillance, cybersecurity — face lighter requirements.

The paper also named market-wide risks. Generative AI could produce fraudulent statements, misleading news and deepfakes to manipulate prices; the suggested control is watermarking AI content. Concentration among a few generative AI providers could create systemic risk; firms would report their vendors, and dominant providers could face monitoring as critical service providers. Herding — firms using similar models and data making similar decisions — could get circuit breakers and collusion audits. Runaway AI flaws spreading across markets could get stress tests, circuit breakers and kill switches.

Is a human in the loop enough?

Pandey himself says no, or at least not necessarily. Speaking at the Global Fintech Fest in September, he said "humans can also overtrust machines" and that oversight "must be competent, empowered and accountable." He added: "an AI-generated alert is not a finding." A named human signature does not guarantee judgment. SEBI's own chairman has flagged the gap.

What is SEBI deploying itself?

The regulator is building its own tools. SEBI's supervisory system SUDARSAN scans public social media content for investment fraud and has flagged more than 20,000 instances of fraudulent content since it began operating in November 2025 — roughly 2,000 flagged items per month over its first ten months. SEBI is demonstrating the tool at international regulatory forums.

On cybersecurity, Pandey cited rising attacks on financial institutions, crypto platforms and digital fraud involving social media influencers. SEBI introduced a structured incident response mechanism in May, aligned with the Financial Stability Board's reporting template. That same month, it directed all regulated entities to overhaul cybersecurity measures amid concerns over Anthropic's Claude Mythos model.

On quantum computing, SEBI is inventorying where markets use cryptography to assess vulnerabilities. Pandey said markets expect these risks to sharpen between 2032 and 2035 — seven to ten years out — and must prepare in advance.

SEBI has also formed a working group to make the Innovators Growth Platform more attractive to startups, particularly deep-tech and biotech firms that take long to reach profitability. Pandey conceded the platform has not gained the desired traction.

The gap to watch: a chairman's speech is not a circular. Until SEBI publishes the guidelines with dates, scope and penalties, firms have a December 2024 liability rule and a consultation paper — not final AI regulation.

via medianama.com (Source)

Filed under

  • ai-governance
  • sebi
  • regulatory-compliance
  • human-oversight
  • capital-markets
Share this article:

More from Daniel Okafor

Daniel Okafor

Show full bio

Staff writer covering technology and product news in enterprise information technology and the leaders who run at CIO Monitor.

209 articles

The newsroom digest

One email a day with the stories that matter. No noise.

By subscribing you agree to the privacy policy.

Cross-references

‹ Previous articleNext article ›