This report presents a practical policy architecture for West Bengal to move from the announcement of AI initiatives to the creation of a credible, institutionalised state AI capability. It examines West Bengalโs current position, benchmarks it against the approaches being adopted by other leading Indian states, and proposes a governance, regulatory, economic and implementation framework for 2026โ2029.
The proposed architecture combines economic development with strong public-interest safeguards. It recommends a high-level AI Taskforce, an operational AI Cell, departmental AI Risk Owners, independent assurance, risk-tiering of government AI systems, mandatory AI Impact Assessments for high-risk applications, a public State AI Registry, Bengali-first citizen services and explicit protections against algorithmic bias, inappropriate surveillance, privacy breaches and AI-enabled electoral disinformation.
Implementation is organised around a 100-day institutionalisation programme followed by a 36-month roadmap, moving from governance and legal foundations to pilots, statewide scale-up and institutional consolidation. Proposed 36-month targets include AI Risk Owners across all major departments, AI-enabled civic services in all 23 districts, โน2,000+ crore in new AI/data-centre investment, training of 25,000+ government employees, Bengali-language AI interfaces across 50+ government services and independent auditing of all high-risk systems before deployment.
๐ ๐ฒ๐๐ต๐ผ๐ฑ๐ผ๐น๐ผ๐ด๐ถ๐ฐ๐ฎ๐น ๐ฑ๐ถ๐๐ฐ๐น๐ผ๐๐๐ฟ๐ฒ: The report itself deliberately excludes Quantum Tiger and Corvus from its policy evidence, recommendations and cited sources. This was intentional: the document is designed to stand on verifiable public evidence, including government documents, national frameworks and peer-state policies, rather than unpublished commercial material. However, Quantum Tiger did use Corvus as an internal AI research and analytical engine in the preparation of this report. Corvus therefore contributed to the research and analytical process, while neither Corvus nor Quantum Tiger is presented as a source of policy evidence or as part of the reportโs substantive recommendations.
The resulting framework is deliberately positioned not as a final policy, but as a working architecture for expert review, costing, legal vetting and public consultation. Its North Star is simple: every significant AI system used by the State should be visible, classified, testable, accountable, contestable and improvable.