SAHI: India's National Framework for Responsible AI in Healthcare
India is set to launch the Strategy for AI in Healthcare for India (SAHI), a national guidance framework to support responsible AI adoption across states, regulators, and private programs. Announced in New Delhi by Minister of State for Health Anupriya Patel, SAHI will be unveiled at the AI Impact Summit 2026 on Thursday.
The Minister reaffirmed a simple idea with big implications: for India, AI is "All Inclusive" - focused on equitable and trusted care for a Viksit Bharat. She noted that AI is already integrated across the health sector, from disease surveillance and prevention to diagnosis and treatment.
What SAHI could mean for healthcare teams
While full details are pending, national guidance of this kind typically sets clear expectations for safety, accountability, and access. Healthcare leaders should anticipate direction in areas like clinical validation, data protection, and real-world monitoring.
- Governance and accountability: Defined clinical owners, oversight committees, and escalation paths for AI incidents.
- Clinical safety and validation: Evidence before deployment, bias checks, and ongoing performance monitoring in real settings.
- Data protection and consent: Privacy-by-design, de-identification where possible, and transparent consent flows.
- Transparency: Model documentation that clinicians can review, and clear patient-facing disclosures when AI supports care.
- Interoperability: Use of standards (e.g., FHIR, SNOMED CT) to reduce vendor lock-in and support data quality.
- Procurement and due diligence: Risk classification, evidence packs, and service-level commitments tied to safety outcomes.
- MLOps and monitoring: Drift detection, audit trails, incident logs, and a safe rollback or "kill switch."
- Workforce readiness: Training for clinicians, nurses, and ops teams; updated SOPs; human-in-the-loop checkpoints.
- Equity and access: Support for local languages and low-resource settings; ongoing impact assessments.
Immediate steps for hospitals and health programs
- Create an AI inventory: list every model/tool in use, its intended clinical purpose, and risk level.
- Stand up an AI governance board with clinical, data, legal, and patient safety representation.
- Adopt a pre-deployment evaluation protocol (clinical validation, bias, security, usability) and require vendor evidence.
- Establish patient consent and communication templates for AI-assisted care.
- Integrate with your EHR/LIS/PACS using standards; avoid one-off data pipes that break at scale.
- Define success metrics per use case (diagnostic accuracy, time-to-diagnosis, false alarm rate, staff workload).
- Run time-boxed pilots in controlled sites; document outcomes and decision gates for scale-up or sunset.
- Set post-deployment monitoring: drift checks, incident reporting, and periodic re-validation.
- Train frontline teams on indications, limits, and fail-safes; make it easy to override or escalate.
What states and regulators can prioritize
- Publish risk-based guidance aligned with SAHI and enable a regulatory sandbox for promising solutions.
- Stand up a shared registry for approved AI tools, known risks, and post-market performance signals.
- Offer procurement templates and evidence checklists to reduce variability and speed safe adoption.
- Support deployments in rural and resource-limited settings; fund evaluations that include equity metrics.
- Coordinate data standards and interoperability requirements across public programs.
For context on good practice in this space, see the WHO's guidance on ethics and governance of AI for health. It outlines principles for safety, transparency, and accountability that can inform implementation while SAHI rolls out. WHO: Ethics and governance of AI for health
Bottom line
SAHI signals a clear direction: AI must improve outcomes, protect patients, and build trust. As the framework is released, assign owners, map your pilots to its requirements, and be ready to show evidence - clinical, operational, and ethical.
For hands-on adoption ideas across surveillance, diagnosis, and treatment, explore AI for Healthcare. For public health leaders and regulators planning policy rollouts, the AI Learning Path for Policy Makers can accelerate consistent, responsible implementation.
Your membership also unlocks:
