Introduction:
As the use of artificial intelligence (AI) and machine learning (ML) in financial markets accelerates, regulators world-wide are grappling with how to encourage innovation while safeguarding market integrity. In India, the Securities and Exchange Board of India (SEBI) has taken a proactive step by proposing a set of guiding principles for the responsible deployment of AI/ML in securities markets. By launching a structured consultation process and assembling an expert panel, SEBI aims to strike the right balance between fostering technological advancement and protecting investors from unintended risks.
1. Background
– Over the last decade, AI/ML applications—ranging from algorithmic trading and portfolio optimization to fraud detection and customer service chatbots—have gained traction among brokers, fund managers and trading platforms in India.
– While these technologies can enhance market liquidity, reduce transaction costs and improve risk management, they also carry potential downsides: biased decision-making, lack of transparency, model over-reliance, data privacy breaches and systemic vulnerabilities.
– Recognizing these challenges, SEBI announced in June 2025 that it would develop non-binding “guiding principles” to ensure that AI/ML tools are used ethically, safely and in a manner that sustains market confidence.
2. Why AI/ML in Securities Markets?
– Enhanced Efficiency: AI-driven analytics can scan vast datasets in real time, allowing trading firms to identify patterns, execute orders and rebalance portfolios at speeds far beyond human capability.
– Improved Risk Management: Machine learning models can detect emerging market anomalies, potential liquidity crunches or counterparty exposures, empowering firms and regulators to respond more swiftly.
– Cost Reduction: Automating routine tasks—such as Know Your Customer (KYC) checks, compliance reporting and customer inquiries—helps firms reduce overheads and focus human expertise on higher-value activities.
3. SEBI’s Expert Panel and Consultation Process
– To craft these principles, SEBI has constituted an Expert Advisory Committee (EAC) on Technology in Securities Markets, comprising academics, data scientists, senior executives from exchanges and market intermediaries, and legal and compliance professionals.
– The EAC will be co-chaired by SEBI’s Executive Director for Market Infrastructure and a distinguished professor of computer science. It will meet monthly over the next four months.
– SEBI has published a consultation paper on its website and invited written comments from stakeholders—including stock exchanges, depositories, registered brokers, asset management companies, technology vendors and investor associations—until August 30, 2025. Two public webinars and sector-specific roundtables are slated for July.
4. Proposed Guiding Principles for Responsible AI/ML Use
The consultation paper outlines an initial set of eight principles, each accompanied by illustrative examples and suggested implementation best practices:
1. Accountability and Governance
– Senior management must own the oversight of AI/ML initiatives. Clear roles and responsibilities should be defined for model development, validation and monitoring.
2. Transparency and Explainability
– Participants should document model logic, data sources and key assumptions. Where feasible, algorithms must be explainable to internal stakeholders and, if required, to regulators or investors.
3. Data Governance and Privacy
– Firms must ensure data quality, integrity and lineage. Personal data used in training or inference must comply with relevant privacy laws and be protected against unauthorized access.
4. Fairness and Non-Discrimination
– AI/ML systems must be tested for bias across demographic groups, trading venues or asset classes. Remedial measures should be implemented to mitigate unfair outcomes.
5. Model Risk Management
– Robust validation frameworks, back-testing protocols and scenario-analysis exercises must be in place. Models should be stress-tested under extreme but plausible market conditions.
6. Human Oversight and Intervention
– Automated decision-making systems must include “kill switches” or manual override mechanisms. Periodic human reviews are required to assess system performance.
7. Cybersecurity and Operational Resilience
– AI/ML pipelines and supporting IT infrastructure must adhere to SEBI’s cyber-security guidelines and industry-standard controls to prevent intrusions, data leaks or denial-of-service attacks.
8. Continuous Monitoring and Incident Reporting
– Real-time monitoring systems should flag anomalous behavior. Material incidents—such as large trading losses attributable to AI-driven errors—must be reported to SEBI promptly, with root-cause analyses and remediation plans.
5. Global Benchmarking and Best Practices
– SEBI’s proposed framework draws heavily on international initiatives, including:
• The International Organization of Securities Commissions (IOSCO) recommendations on technology risk management;
• The Organisation for Economic Co-operation and Development (OECD) Principles on Artificial Intelligence;
• Regulatory sandboxes and thematic reviews conducted by the U.K. Financial Conduct Authority (FCA) and the U.S. Securities and Exchange Commission (SEC).
– By aligning with these global standards, SEBI seeks to ensure interoperability, encourage cross-border market participation and prepare Indian market participants for evolving international compliance requirements.
6. Next Steps and Timeline
– Public Consultation: Comments due by August 30, 2025; webinars in July and August.
– Draft Finalization: The Expert Advisory Committee will synthesize feedback and publish a revised draft by October 2025.
– Adoption and Implementation: SEBI aims to issue the final set of guiding principles by December 2025. Market participants will be granted a transition period—likely six to twelve months—to align their processes, update policies and conduct necessary staff training.
– Ongoing Review: SEBI will monitor the adoption of the principles and may conduct periodic thematic reviews or “deep-dive” inspections, adjusting the framework over time to reflect emerging technologies and lessons learned.
Key Takeaways
• SEBI is proactively formulating non-binding guiding principles to promote ethical, transparent and resilient use of AI/ML in India’s securities markets.
• The proposed framework rests on eight core pillars—including accountability, explainability, data governance and model risk management—mirroring global best practices.
• A structured consultation process (closing August 30, 2025) and staged implementation (finalization by December 2025, with a transition period thereafter) aim to balance innovation with investor protection.
Frequently Asked Questions
1. Who will need to comply with these AI/ML guiding principles?
All market participants—stock exchanges, brokers, asset managers, depositories and technology vendors—using AI/ML tools in trading, analytics, customer interfacing or back-office operations.
2. Are these SEBI guidelines legally binding?
No. They are intended as non-binding “guiding principles” designed to shape industry practices. However, SEBI expects firms to adopt them in good faith and may consider formal rule-making if market conduct issues emerge.
3. How can stakeholders provide input during the consultation?
Interested parties can submit written comments through SEBI’s online portal, register for public webinars (details on SEBI’s website) or participate in sector-specific roundtable discussions scheduled for July 2025.
