NETRA: An Open, Verification-First Observatory of AI Harms, Law and Accountability in India

India has no central register of harms caused or enabled by artificial intelligence, and no single map of the laws and authorities that may answer them. NETRA is an open, verification-first observatory that records AI-linked harms in India, reads each one against Indian law, and routes it to the bodies that can act. Each case is traced through a 14-stage lifecycle, and risks are coded to the MIT AI Risk Repository taxonomies. Release v0.3.1 holds 22 incidents reviewed by the author, 28 Indian legal instruments (17 verified against official texts, 11 partly verified), 18 risks quoted from Indian government frameworks, 28 measures, 6 governance gaps, a directory of 391 Union and state bodies and 12 AI model cards. Version 3 separates reported from confirmed AI involvement: AI's role is confirmed in 10 of 22 cases, suspected in 9 and not established in 3. This working paper describes the method, the verification rules and the dataset, reports descriptive patterns only, and sets out the limits of a small, single-maintainer, English-language record. Research material, not legal advice.