India’s Ministry of Electronics and Information Technology (MeitY) has put forward a new Data Empowerment and Protection Architecture (DEPA) specifically tailored to facilitate access to large-scale datasets needed for AI training, while prioritizing citizen privacy and regulatory compliance.
- Consent-driven, privacy-centric data sharing for AI training
- Utilizes confidential clean rooms and differential privacy techniques
- Includes mechanisms for continuous model oversight and fairness checks
What happened
MeitY has proposed a detailed consent-based framework named DEPA for AI training that builds upon India’s existing data-sharing architecture. It enables organizations like banks, hospitals, and government agencies to securely share anonymized data with AI developers without releasing the underlying data directly. The framework introduces technical pillars such as confidential clean rooms to provide secure computing environments, differential privacy to prevent identification of individuals from aggregated data, and machine-readable electronic contracts to enforce conditions governing data use.
Under this framework, datasets are registered on a data exchange platform with metadata and cryptographic identifiers for traceability. AI developers, termed Training Data Consumers (TDCs), request access to this data for specific use cases, which is granted only upon verification against defined policies. Model training occurs within a secure environment to protect raw data. Post-training, models undergo independent assessments for bias, security, and explainability before being approved for deployment.
Why it matters
India’s AI landscape requires access to diverse and large-scale datasets to develop robust and contextually relevant AI models. However, concerns about privacy, regulatory compliance, and data ownership have limited effective data sharing. MeitY’s framework introduces a tested approach to balance these concerns by allowing data providers to maintain control and ensure privacy while enabling innovation. This could accelerate AI advancements across startups, enterprises, and research institutions in India.
Furthermore, with increasing global scrutiny on AI ethics and governance, the framework’s emphasis on continuous oversight, fairness, and auditability supports responsible AI development. The initiative also reflects the Indian government’s broader commitment to shaping AI policy, seen concurrently in regulatory consultations by telecom agencies and upcoming government papers addressing AI safety and security.
What to watch next
Stakeholders should monitor the finalization and rollout of this framework, including the registration and operationalization of the proposed data exchange platforms and clean room environments. How startups and enterprises adopt these mechanisms will be a critical indicator of the framework’s effectiveness in unlocking AI innovation while preserving privacy.
Additionally, upcoming regulatory feedback rounds, such as TRAI’s draft AI guidelines and MeitY’s planned consultation paper on AI safety, will provide further clarity on India’s evolving AI governance ecosystem. The integration and compatibility of DEPA with other sectoral regulations and data protection laws will also be pivotal for its sustained impact.