Overview
- TRAI released the consultation draft Friday requiring operators to adopt board-approved AI policies, create AI governance and safety committees, and keep central inventories that classify each AI system as high, medium or low risk.
- The rules mandate pre-deployment validation, conformity and impact assessments covering accuracy, robustness, explainability, fairness and model-drift resilience before any AI system can be used.
- Critical and high-severity AI incidents must be reported to TRAI and operators must maintain central databases of AI incidents and consumer complaints, though the draft does not define severity tiers, reporting formats or deadlines.
- Operators must do supplier due diligence, obtain vendor disclosures about embedded AI and include enforceable audit and incident-management clauses in contracts because the draft makes the operator accountable for third-party AI outcomes.
- Stakeholders can comment until 27 November 2026 and TRAI says it will prescribe the missing procedural and enforcement details later, while enforcement for breaches will rely on existing telecom laws and licence conditions.