India's AI Approach: Innovation First, Guardrails When Needed
India's position on AI is clear: keep the door wide open for innovation, and step in with regulation when it's truly required. At the release of the IndiaAI governance guidelines, IT Secretary S. Krishnan said the government wants AI to deliver maximum public benefit, while staying ready to legislate if harms emerge.
"If we believe that the priority needs to be for innovation, regulation is not the priority today... if the need arises for legislation or regulation, the government will not be found wanting," he said. The report backs a human-centered approach and supports the current stance of not rushing new laws.
What the new AI governance guidelines emphasize
- Trust
- People-first approach
- Innovation over restraint
- Fairness and equity
- Accountability
- Clear disclosures and explanations for users and regulators
- Safety, resilience, and sustainability
Prepared by a sub-committee chaired by IIT Madras Professor B. Ravindran, the guidance fine-tunes measures the government has already been following. The intent: encourage progress while protecting citizens from obvious harms.
Short-term actions recommended
- Stand up key governance institutions for AI oversight and coordination.
- Develop an India-specific AI framework aligned to local needs and public service delivery.
- Identify legal amendments needed to address immediate gaps.
- Expand access to compute, data, and shared infrastructure for AI projects.
- Increase access to AI safety rules and practical guardrails for implementers.
Medium-term plan
- Publish common technical and process standards for public-sector AI.
- Amend laws and regulations where needed to reflect AI risks and usage.
- Operationalise an AI incident reporting and response system.
- Pilot regulatory sandboxes to test high-impact use cases with safeguards.
Ongoing commitments
- Build capacity across ministries and industry; set and update standards.
- Review and refine the government's framework as technology and risks evolve.
- Draft new laws when new capabilities or harms make them necessary.
Principal Scientific Advisor Ajay Sood called for ministries and industries to form working groups that look at both safeguards and new applications. Additional Secretary Abhishek Singh noted the recommendations draw on public consultation, including about 650 comments.
What you can do now in your department
- Map your AI pilots and planned use cases. Flag those affecting benefits, eligibility, or enforcement for added review.
- Designate an AI point of contact for incident reporting and risk assessment.
- Adopt a simple pre-deployment checklist: purpose, data sources, model risks, human oversight, explainability, and grievance redressal.
- Require vendors to provide plain-language model disclosures and decision explanations that your users and regulators can understand.
- Run a bias and equity review for any model influencing citizen outcomes. Document mitigation steps.
- Propose a sandbox for high-stakes use cases to test with limited scope, clear metrics, and rollback plans.
- Plan training for your teams on AI fundamentals, risk management, and procurement standards.
Resources
- IndiaAI (official portal)
- Ministry of Electronics and IT
- AI courses by job role (for public-sector upskilling)
The bottom line: keep pushing for high-utility AI in public services, with people-first safeguards. Build capacity, standardize what works, and be ready with regulation when real risks show up.
Your membership also unlocks:
