TRAI draft would make telecom firms govern their AI systems
TRAI's draft guidelines ask telcos to adopt board-approved AI policies, risk-rank every AI system and report serious incidents. Comments close 27 November.
The Telecom Regulatory Authority of India (TRAI) released draft guidelines on 30 October 2026 on the responsible use of artificial intelligence in telecom networks and services, according to PSU Watch and an IANS report carried by SocialNews.XYZ. Operators would need a board-approved AI policy, a risk class for every AI system and a duty to report serious AI incidents to the regulator. Written comments are due by 27 November 2026.
What TRAI has proposed
The draft applies to any entity that holds a licence under the Telegraph Act, 1885, or an authorisation under the Telecommunications Act, 2023, and that uses or plans to use AI in its networks or services. TRAI says the aim is to apply the India AI Governance Guidelines consistently in telecom and to "increase clarity, predictability and accountability" for the sector.
The central idea is accountability for outcomes. An operator would answer for what its AI does whether it built the system, bought it from a vendor, or used a mix of both. Buying a model from a third party would not move the responsibility elsewhere.
The building blocks of the draft
According to PSU Watch's summary, the draft asks each entity to do six things:
- Adopt an internal AI policy approved by its board or an equivalent body. It would cover risk classification, deployment approvals, incident management, human oversight, data governance and consumer grievances.
- Sort every AI system into high, medium or low risk, based on how important it is to network operations, how much it affects financial and operational results, and how it touches consumers.
- Set up an AI Governance and Safety Committee led by a senior executive called the AI Governance and Safety Officer. The committee would review systems at least once a financial year and could order changes or decommissioning.
- Run validation, conformity and impact assessments before deploying a new or changed system, checking accuracy, robustness, explainability, fairness and resilience to "model drift" (when a model slowly gets worse because real-world data changes).
- Keep an inventory of active, inactive and retired AI systems with each one's source, purpose and risk class. Systems missing from the inventory could not be used.
- Rate incidents as critical, high, moderate or low, and report critical and high ones to TRAI in a prescribed format and timeframe.
Operators would also have to keep central records of AI incidents and of consumer complaints about AI-related harm, and do due diligence before buying AI products.
Why the regulator is moving now
TRAI's stated reasoning is that AI can make networks more efficient but also brings unintended model behaviour, bias and attacks designed to fool models, any of which could affect critical infrastructure.
What is confirmed is the content of the draft and the comment deadline. What is not yet known is how strict the reporting timelines will be, because the reports say they will follow a "prescribed" format and timeframe that is not spelled out in the coverage we could open. The final rules may also change after consultation.
What it means for telcos, vendors and users
For operators, the practical task is an audit: list every AI tool in use, from fraud detection to network optimisation and customer chatbots, and decide which are high risk. Vendors selling AI to telecom companies should expect more questions on testing, explainability and how their models behave over time.
For customers, the draft promises a formal route for complaints when an AI decision causes harm, such as a wrongly blocked connection. It does not yet say what remedy a customer would get.
Try the planner below with your own numbers.
What to watch next
Comments close on 27 November 2026. Watch for the final guidelines, the exact incident-reporting timelines, and whether other regulators such as the RBI and SEBI issue similar sector-specific rules under the same national framework.
Key takeaways
- TRAI published draft guidelines on responsible AI in telecom on 30 October 2026, with comments due by 27 November.
- Telecom operators would be accountable for AI outcomes even when the system comes from a third-party vendor.
- Each AI system would carry a high, medium or low risk class and sit in a mandatory inventory.
- Critical and high-severity AI incidents would have to be reported to TRAI.
- The draft is not final, and the reporting timelines are not yet clear from published coverage.
Sources
- [PSU Watch: TRAI proposes board-approved AI policies, mandatory reporting of serious incidents by telcos](https://psuwatch.com/newsupdates/trai-proposes-board-approved-ai-policies-mandatory-reporting-of-serious-incidents-by-telcos)
- [SocialNews.XYZ (IANS): TRAI releases draft guidelines on responsible use of AI in telecommunication networks](https://www.socialnews.xyz/2026/10/30/trai-releases-draft-guidelines-on-responsible-use-of-ai-in-telecommunication-networks/)
- [Inc42: MeitY floats framework to unlock data for training AI models (mentions the TRAI consultation)](https://inc42.com/buzz/meity-floats-new-privacy-led-framework-to-unlock-data-for-training-ai-models/)
Tags: TRAI AI governance Telecom Regulation AI safety