Tamil Nadu Safe & Ethical Artificial Intelligence Policy, 2020

Tamil Nadu has long presented itself as a leader in e-governance. The policy's own introduction describes heavy investment in IT infrastructure, end-to-end citizen-service applications and a start-up ecosystem, and notes that the Chief Minister had recently announced a Centre of Excellence in Emerging Technologies (CEET) under the Tamil Nadu e-Governance Agency (TNeGA), mandated to help departments use AI, the Internet of Things, blockchain, drones and AR/VR. By then CEET had begun building solutions for health, education, rural development, agriculture and land registration, and several image-recognition tools were ready to roll out.¹

The document sets out the reasoning for a state policy. It argues that AI systems have features that make them hard to govern, including opacity, diffuse development and difficulty in assigning responsibility, and that poorly designed systems used in law enforcement, banking or healthcare can interfere with the right to life or freedom. It says government has two conflicting roles, as a user of AI to deliver services and as a regulator keeping the technology aligned with human values, and states its goal as harnessing AI for the public good while keeping it safe and ethically compatible with human values.¹

The policy appears under the "IT Policies, Acts and Rules" listing on the Tamil Nadu IT Department's website, next to the state's 2020 Blockchain Policy, its 2020 Cyber Security Policy and its 2022 Data Policy.² That grouping matters, because the AI policy's certification mechanism depends on the blockchain infrastructure.

The policy lists six goals: AI that is inclusive, free of bias, fair, equitable, ethical and transparent; fast and transparent public services through AI-driven e-governance; better governance workflows; awareness and adoption of AI across the Legislature, Judiciary and Executive; research and development by academia, start-ups and enterprises; and an ecosystem of open data access. Its primary objectives include a framework for inclusive, safe and ethical use of AI in government, guidelines to evaluate an AI system before public use, a self-sustaining AI community and skilling in Tamil Nadu, access to open data, data models and computing resources, a regulatory sandbox, and investment in AI research and development.¹

Scope. The policy applies to authorities and bodies owned or controlled by the Government of Tamil Nadu or receiving aid from it, to cooperatives, trusts, societies, public sector undertakings and boards whose composition or officers are controlled by the state, and to the state's partnerships and joint ventures.¹

TAM-DEF. Every AI system is to be evaluated before roll-out against six challenges, taken from academic literature that the policy cites: Transparency and audit (explainability and decision audit trails), Accountability and legal issues (who is liable when a system decides autonomously), Misuse protection, Digital divide and data deficit (communities with poor data are left behind), Ethics (privacy and data protection, and human and environmental values) and Fairness and equity (no gender or racial bias and no social profiling, especially in law enforcement and fraud detection).¹

DEEP-MAX. The seven-parameter scorecard turns these challenges into a points-based rating: Diversity, Equity and fairness, Ethics, Privacy and data protection, Misuse protection, Audit and transparency, and cross-geography and cross-society performance. TNeGA is to help procuring entities assess whether a system meets the scorecard. Scoring begins as binary (1 or 0 per parameter), with TNeGA aiming to introduce a 0-10 scale for each parameter and to partner with research institutions, standard-setting bodies and other governments to build test datasets.¹

The policy turns the framework into procurement rules. Before buying any AI solution, a department must consult CEET at TNeGA, ensure the system carries a satisfactory DEEP-MAX score, and encourage periodic re-calibration of scores, suggested at six months at most because AI systems are self-learning. The Government of Tamil Nadu, or agencies acting for it, retains the right to set and revise acceptability criteria and to reject or recall an application that does not meet them.¹

The document proposes an AI Certification, Transparency & Scorecard (ACTS) Blockchain, to be designed by TNeGA and made available after the state blockchain backbone, called Nambikai Inayam in the policy, is rolled out. All DEEP-MAX scores for AI used in the public domain are to be stored and regularly updated there, and the ACTS Blockchain is described as a universal and publicly viewable ledger that would also certify whether training data met diversity and equity criteria. One illustration given is a face-recognition system for apprehending criminals: a blockchain record of changes to the criminal image database, with authorisation details, would make substitution or tampering traceable. Until the ACTS Blockchain is live, TNeGA is to devise an alternative mechanism for storing the scores.¹

Nambikkai Inaiyam is TNeGA's blockchain-as-a-service infrastructure, developed in line with the Tamil Nadu Blockchain Policy. TNeGA's page for it says it has secured 27 categories of e-Sevai certificates and over two crore documents, and does not mention an AI scorecard or an AI policy.⁴

Oversight rests with a Safe & Ethical AI Monitoring Committee headed by the Chief Secretary, with secretaries and senior officers of selected departments, including the law department, and AI and policy experts from academic and research institutions. The IT Department coordinates overall implementation and may issue guidelines and revisions.¹

A large part of the policy is promotional. It encourages every department to use AI where appropriate, with TNeGA helping to identify opportunities, and proposes a set of enabling measures:¹

• Expert groups of academics, practitioners and departmental experts, to hold conclaves, identify use cases, draft sector-specific guidelines and study the real-world impact of deployed systems.
• Open AI challenges for students, teachers, start-ups and enthusiasts, with TNeGA helping to commercialise winning solutions on a joint-ownership basis for intellectual property.
• Collaborative core and applied AI research centres focused on AI for governance.
• Open data, through an open data task force headed by TNeGA's Chief Executive Officer, a data platform and departmental data catalogues, with privacy and confidentiality respected and Union and state law observed.
• An AI sandbox giving start-ups and individuals access to compute and data without the risk of failure.

The policy also calls for awareness and training for four audiences. For citizens it proposes AI content in school and college curricula, large-scale youth skilling and public awareness programmes. For the executive, orientation sessions for officers. For the Legislature, primer courses and exchange programmes so that law-makers can legislate effectively. For the Judiciary, a programme, in consultation with the High Court, to educate judicial officers about AI and its legal implications.¹

The policy is an executive document. Several limits are apparent from its text. It binds the state's own agencies and bodies they control or fund, not private developers, and it sets out no penalties or enforcement powers beyond the right to reject or recall non-compliant applications. Its scorecard depends on the availability of suitable test datasets, which the policy says TNeGA would seek to develop with partners, and its public certification layer depends on infrastructure that the policy describes as still to be designed.¹

The sources reviewed for this entry do not show whether the ACTS Blockchain or a continuous 0-10 scale has been implemented, whether the Monitoring Committee has published any reviews, or how many AI procurements have gone through the DEEP-MAX process. TNeGA's website, as fetched for this entry, links no ethical-AI policy page. This entry therefore describes the policy as designed and does not assess how it has operated. No reputable external criticism of the policy was located in the sources available for this entry.

The policy's emphasis on pre-deployment testing and public scorecards is notable against the national picture. NITI Aayog's Responsible AI principles for India came in 2021, and the national AI governance guidelines in 2025. Tamil Nadu's document was issued before either, and it concentrates on public-sector procurement, where the state is both buyer and regulator.

As of 15 October 2026 the policy remains posted on the Tamil Nadu IT Department's policy page, and I found no notice repealing or superseding it.² The state has since moved on to larger AI programmes. On 13 January 2026 the Tamil Nadu government signed an MoU with Sarvam AI for a Sovereign AI Park near the IIT Madras Research Park in Chennai, with an initial investment reported at Rs 10,000 crore, a dedicated Institute for AI in Governance and plans for Tamil-first foundation models; MediaNama noted that the amount was almost equal to the Union government's IndiaAI Mission outlay and that the state had shared no timeline.⁵

On 12 August 2026 TNeGA published an announcement that Tamil Nadu had launched the TN AI Mission, announced as part of the Tamil Nadu Budget 2026-27, to embed AI-driven solutions across public services, with priority sectors of healthcare, agriculture, education and citizen services and possible extension to urban administration and social welfare. The announcement states no budget figure and does not mention the 2020 policy.³