A risk assessment and classification framework customised to India's social, economic and cultural context addresses the main risk categories (malicious uses such as deepfakes and data poisoning, bias and discrimination, transparency failures, systemic risks, loss of control, national security) with particular protection for vulnerable groups such as children exposed to engagement-maximising recommenders and women targeted by deepfakes. A national AI incidents database, centralised but able to query federated local databases kept by authorised entities and sectoral regulators on a standard schema, collects empirical data on AI harms, extends to securely held classified threat intelligence, leverages CERT-In mechanisms, encourages confidential self-reporting without the threat of penalties, and feeds a structured loop from reports to threat analysis to oversight.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.