Indian Artificial Intelligence Stack
India · Framework · AI Standardisation Committee, Department of Telecommunications
Our summary
This document proposes a comprehensive framework for developing an India-specific AI architecture divided into six layers. It applies to government agencies, private sector stakeholders, manufacturers, service integrators and cloud service providers working to establish uniform standards and enabling conditions for AI deployment across sectors. The framework aims to remove barriers to AI adoption and create an integrated environment supporting inclusive growth.
Written by the Index in our own words. It is not legal advice; the official text is the authority.
Inside this policy
Every section of this policy in order. Our notes summarise each one in plain language and do not quote the official text, which is the authority.
Drafted by the Index An automated summary of the official text, not yet reviewed by a person.Reviewed Checked by a person against the official text.
11 sections
Jump to a chapter
1: introduction
Artificial intelligence is a key technology of the Fourth Industrial Revolution with potential to address major national challenges and support development across sectors including clinical services, farming, learning systems, financial services and public administration, offering benefits such as helping identify human biases in decision-making, enabling smart business solutions, automated services, and new opportunities including autonomous vehicles and digital assistance. While AI offers significant economic and societal benefits, safeguards and oversight are needed to prevent harm, and India must invest in AI innovation to remain competitive. This transformation will require new architectural designs and upgraded technologies to support real-time decision-making.
2: AI’s IMPACT ON SOCIETY
AI will have transformative social, economic, legal, political and regulatory effects, with social impacts including integration into daily life through virtual assistants and chatbots. Economic benefits are expected across manufacturing, professional services, financial services and retail through automation and innovation, with specific gains identified in healthcare diagnostics and treatment, agriculture crop yield and pest detection, smart mobility autonomous systems, retail personalised suggestions, manufacturing smart production and predictive maintenance, energy system modelling and grid management, smart cities traffic and security, education personalised learning and intervention prediction, and banking and financial services customer assistance, operations automation and fraud detection. India ranks third globally in AI manufacturing implementation, with 19 percent of companies significantly using AI.
3: definition of AI
This section presents multiple definitions of artificial intelligence from various researchers and organizations, ranging from the ability of machines to exhibit human-like intelligence to the capability to learn from data and perform complex tasks through flexible adaptation.
4: need for an indian AI stack
India faces barriers to AI adoption including insufficient research expertise, lack of intelligent data, high costs, privacy gaps, and fragmented implementation; risks include biased training data, opaque proprietary algorithms, and unmonitored automated decision-making creating unfair outcomes and ecosystem-wide threats. A standardised Indian AI stack covering data protection, federation, minimisation, algorithms, interfaces, monitoring, audit, logging, privacy, ethics, and digital rights is needed to ensure transparent, ethical architecture with unbiased data, proper storage, industry-wide standards, developer accountability, and independent agency oversight that controls risks without deterring innovation.
5: proposed indian stack
The proposed Indian AI Stack comprises six layers: a data exchange layer with consent-based data access and open APIs; an application layer for software services with transaction tracking and audit functions; a compute layer for AI/ML analytics using techniques including natural language processing, deep learning and neural networks; a storage layer managing hot data on fast media and cold data on slower media for archival; an infrastructure layer implementing data federation, encryption, minimisation and multi-cloud controller architecture with multi-tenant computing and multi-layer storage; and a cross-cutting security and governance layer applying cybersecurity protections across all layers against data compromise, integrity attacks and cloud access threats through encryption and cryptographic support. Data controllers determine purposes and means of personal data processing, with protection mechanisms required to address privacy risks from mining large datasets.
6: layers of the indian AI stack
The Indian AI Stack comprises five layers: the infrastructure layer manages cloud data controllers, traffic, segregation, encryption, minimisation and federation while ensuring lawfulness, fairness, transparency and data integrity; the storage layer classifies data as hot, warm or cold and maintains secure, scalable storage with durability mechanisms; the compute layer collects and analyzes data using machine learning, deep learning and natural language processing, transporting processed data via ingestion tools and generating insights through open algorithm frameworks; the application layer hosts software and services with centralized web-based access, multi-tenant architecture, load balancing, billing tools and audit logging; and the data information exchange layer forms the top interface defining trustworthiness, edge security, ethical standards, digital rights protection and consent frameworks while enabling open API integration, data sharing across vendors and inclusion of partners from public, private and voluntary sectors.
Vertical cross layer
Establishes a security and governance layer spanning all horizontal layers to ensure services are safe, trusted, and protected. The layer mandates implementation of cybersecurity frameworks, certification schemes for critical AI services, data protection governance, encryption methods, digital forensics capabilities, and compliance with privacy legislation.
7: AI stack flowchart
Outlines a fifteen-step process for data flowing through the stack: from initial input through multi-cloud data control, encryption, storage categorization, scaling and analytics, transaction recording and audit logging, to final delivery through the data exchange layer with consent and digital rights verification, and distribution via open APIs.
Annexure, 1
This section lists countries that have developed or are developing national artificial intelligence strategies or programs, from Argentina through Russia, including their specific initiatives such as dedicated funding, task forces, legislation, or named strategies. This section continues listing countries with artificial intelligence strategies or programs, from Saudi Arabia through Uruguay, detailing their respective policy initiatives and governance frameworks for AI development.
Annexure, 2
References
Controls
What this policy requires, forbids, grants or penalises, one control at a time and in our own words. Each control links to the official text, which is the authority. This is not legal advice.
Drafted by the Index Drafted automatically from the official text, not yet reviewed by a person.Reviewed Checked by a person against the official text.
Themes covered 8 of 35
Choose a theme to see only its controls. About the themes
Risk and scope 1 of 4
- Risk classification, not covered
- Prohibited practices, not covered
- Impact assessment, not covered
Data and privacy 2 of 4
- Cross-border data transfer, not covered
- Copyright and training data, not covered
Transparency and records 2 of 4
- Labelling of synthetic content, not covered
- Technical documentation, not covered
Technical safeguards 2 of 4
- Human oversight, not covered
- Testing and evaluation, not covered
Assurance and lifecycle 0 of 4
- Conformity assessment and certification, not covered
- Registration and notification, not covered
- Incident and breach reporting, not covered
- Post-market monitoring and corrective action, not covered
Rights and protected groups 0 of 4
- Automated decision-making rights, not covered
- Children and vulnerable groups, not covered
- Biometric identification and surveillance, not covered
- Whistleblowing, complaints and redress, not covered
Roles and organisation 1 of 6
- Supply chain and provider-deployer duties, not covered
- General purpose and foundation models, not covered
- Training and AI literacy, not covered
- Public sector use and procurement, not covered
- Innovation support and sandboxes, not covered
Enforcement and sectors 0 of 5
- Enforcement, penalties and regulators, not covered
- Sector rules: health, not covered
- Sector rules: finance and insurance, not covered
- Sector rules: employment, not covered
- Sector rules: education, not covered
The full list of controls could not be loaded, so only the first 15 are shown.
4: need for an indian AI stack 1 control
- ObligationApplies to: Public authorityDrafted by the Index
A national AI stack should establish standards for data protection, open algorithms, ethical practices, trustworthiness, monitoring, audit and logging across all sectors.
5: proposed indian stack 6 controls
- ObligationApplies to: Public authorityDrafted by the Index
The Indian AI stack should provide five horizontal layers (infrastructure, storage, compute, application, and data exchange) plus a vertical security and governance layer to deliver secure, trustworthy AI services.
- ObligationApplies to: Public authorityDrafted by the Index
The stack should maintain secure storage environments with data classification, protection, federation, and minimization capabilities.
- ObligationApplies to: Public authorityDrafted by the Index
The stack should ensure legitimacy of backend services and transaction movement through secured gateway services.
- ObligationApplies to: Public authorityDrafted by the Index
The stack should protect digital rights and maintain ethical standards for AI services delivered to customers.
- ObligationApplies to: Public authorityDrafted by the Index
The stack should establish a properly framed consent framework to obtain customer consent for data use.
- ObligationApplies to: Public authorityDrafted by the Index
The stack should enforce use of Government Public Key Infrastructure services for all operations.
6: layers of the indian AI stack 7 controls
- ObligationApplies to: ProviderDrafted by the Index
The infrastructure layer should establish a common data controller across multi-cloud and hybrid cloud environments.
- ObligationApplies to: ProviderDrafted by the Index
The infrastructure layer should implement data federation, encryption and minimization techniques to protect data in cloud storage.
- ObligationApplies to: ProviderDrafted by the Index
The infrastructure layer should monitor and maintain privacy of stored data and ensure lawfulness, fairness and transparency in processing.
- ObligationApplies to: ProviderDrafted by the Index
The infrastructure layer should separate and manage public and private data and develop data sharing structures for public and private data hubs.
- ObligationApplies to: ProviderDrafted by the Index
The storage layer should categorise data as hot, warm or cold depending on access frequency and ensure appropriate storage based on data classification.
- ObligationApplies to: ProviderDrafted by the Index
The storage layer should maintain secure storage, ensure data durability, define archiving and extraction procedures, and establish data locality roles.
- ObligationApplies to: ProviderDrafted by the Index
The compute layer should perform AI and ML analytics using techniques such as deep learning, machine learning and natural language processing.
Annexure, 2 1 control
- ObligationApplies to: Public authorityDrafted by the Index
The Department of Telecommunications has formed a committee to develop standards for artificial intelligence and build an Indian AI stack.
Change history
No changes recorded since we began monitoring this policy.