On October 30, 2026, the Telecom Regulatory Authority of India (TRAI) released draft "Guidelines on Responsible Use of Artificial Intelligence (AI) in Telecommunication Networks and Provision of Telecommunication Services". "Written comments on the draft guidelines are invited from stakeholders by 27.11.2026." The draft says it is meant to ensure "consistent and responsible implementation of the India AI Governance Guidelines in the telecommunication sector".
Who it covers and what it asks
It applies to any entity holding a licence or authorisation to provide telecom services or run telecom networks that "is using or intending to use AI system(s)" in them. Each would need an internal AI policy "approved by the board of directors", filed with TRAI "within seven working days" of taking effect. The policy must include:
- A three-level risk classification: every AI system "shall be classified as high-risk, medium-risk, and low-risk", based on its importance to the network or service, including financial, operational and consumer impact.
- Four incident severities: "critical-severity AI incident, high-severity AI incident, moderate-severity AI incident, and low-severity AI incident."
- Approval before use: "Any AI system shall be used only after approval by an officer" designated for the purpose.
- Human oversight, data governance and a grievance process for AI-related harms reported by users.
Operators would also have to keep an inventory of "all active, inactive and decommissioned AI systems", and "no AI system shall be used unless it is a part of the inventory." They would carry out due diligence on third-party AI, obtain AI disclosures from equipment suppliers, write audit and incident terms into supplier contracts, and run a "conformity assessment" covering "accuracy, robustness, explainability, fairness, drift resilience and human oversight" before deploying or modifying a system. Incidents, "including testing events", would go into a central database using TEC standard 57090:2025 or an equivalent taxonomy. An AI Governance and Safety Committee headed by a senior officer would review deployed systems at least yearly.
Why it matters
The draft puts accountability with the operator: it is "accountable for the outcome of all AI systems used by it, irrespective of whether the AI systems are developed internally, procured/ outsourced from third party". For vendors of AI-enabled network equipment, that will arrive as disclosure requests and contract clauses. Including testing events in the incident database is notable: failures found in evaluation would be recorded, not only those that reach customers.
What remains uncertain. Critical and high-severity incidents go to TRAI "in a prescribed format, manner and timeframe", and none of these is defined in the draft. The thresholds between risk levels are left to each operator's policy. Press coverage describes the draft as consultative, and it may change after comments.
Analysis: Belgavi's Book Store is not a telecom operator, but the inventory rule is a good habit for any business. Hansini could start a one-page list of every AI system the store runs, from the help-desk assistant to the returns classifier, with its supplier, purpose and a risk level, and add a rule that nothing new goes live until it is on the list.
Sources and evidence
TRAI's draft responsible-AI guidelines for telecom, open for comment until November 27, would require a board-approved AI policy filed with TRAI, three risk levels, four incident severities, an inventory of all AI systems, pre-deployment conformity assessment, supplier disclosures, an incident database including test events and a senior governance officer. Reporting timeframes are not yet defined.