ICMR Issues Landmark Ethical Guidelines to Govern AI Use in India’s Healthcare and Biomedical Research
The Indian Council of Medical Research (ICMR) serves as India’s primary authority for governing AI use in healthcare and biomedical research. In March 2023, ICMR released “Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare,” establishing a framework that ensures AI tools in medicine are safe, fair, transparent, and patient-centric. This overview covers how ICMR’s governance will impact India’s healthcare sector.
The 10 Core Ethical Principles
ICMR established 10 foundational principles that guide all AI development and deployment in healthcare:
- Accountability and Liability – Clear responsibility when things go wrong
- Autonomy – Patients and doctors retain decision-making authority; AI assists but doesn’t replace
- Data Privacy – Patient health information is protected and anonymized
- Collaboration – Interdisciplinary cooperation among stakeholders
- Safety and Risk Minimization – All AI tools must be proven safe before clinical use
- Accessibility and Equity – AI benefits must reach everyone, not just wealthy urban populations
- Data Optimization – High-quality, unbiased, representative datasets
- Non-Discrimination and Fairness – AI must work fairly across all populations, genders, and regions
- Trustworthiness – AI solutions must be transparent, reliable, and valid
- Clinical Validation – Rigorous testing and approval before doctors use AI on patients
Key Changes for Healthcare Stakeholders
- For Patients:
- Must receive informed consent before AI is used in their care
- Have the right to know when AI influences their diagnosis or treatment
- Their health data is protected through anonymization and separation from personal identifiers
For Hospitals and Doctors:
- Cannot use new AI diagnostic tools without first proving they work through clinical validation
- Must obtain ethics committee approval before deploying AI systems
Doctors maintain final decision-making authority; AI is a tool, not a replacement
- For AI Developers:
- Must train AI systems on diverse, representative Indian datasets to prevent bias
- Must ensure AI explainability—doctors and patients should understand how decisions are made
- Subject to government medical device regulations if developing diagnostic tools
For Data Protection:
Patient health data used in AI must be anonymized and delinked from personal information
Protects against data breaches and misuse as healthcare digitalization increases
How ICMR’s Governance Works in Practice
- Ethics Committee Review: Before testing any AI tool on patients, developers must present it to an institutional ethics committee. The committee assesses scientific rigor, patient safety, and ethical compliance.
- Medical Device Regulation: AI tools classified as medical devices (diagnostic software, treatment recommendations, etc.) require approval from the Central Drugs Standard Control Organization (CDSCO) before commercial use.
- Global Collaboration: India joined the HealthAI Global Regulatory Network in September 2025, enabling collaboration with countries like the UK and Singapore on AI healthcare standards.
- Living Guidelines: ICMR’s framework is designed to evolve as technology and understanding improve.
Any AI system used in biomedical research or clinical settings must safeguard patient privacy, avoid discriminatory outcomes, and remain fully accountable to regulatory and ethical requirements. Unauthorized or irresponsible use of patient data, deployment of unvalidated AI tools, or failure to provide explainability and oversight may constitute violations of ethical and legal obligations. These guidelines caution developers, researchers, and healthcare institutions that AI must support clinical judgment not replace it and that non-compliance may lead to professional, ethical, or regulatory consequences.
____________________________________________________________
Astrea Legal Associates LLP
Contributed by Urwi Keche , Partner & Naisergi Desai, Trainee
Note: This publication is provided for general information and does not constitute any legal opinion.This publication is protected by copyright.
© 2025,Astrea Legal Associates LLP



