TRAI Notifies New Anti-Spam Rules: What AI Checks, Robocall Declarations and Appeals Mean
TRAI's new anti-spam rules add phased AI-led checks, robocall declarations, caller-ID reporting and clearer complaint appeals for India.
- TRAI's Telecom Commercial Communications Customer Preference (Third Amendment) Regulations, 2026 are dated 18 September and introduce phased, Gazette-linked commencement rather than an immediate anti-spam switch-on.
- The amendment defines an A2P call as an automated voice call initiated without direct human dialling, including autodialling, robocalls and pre-recorded or artificial-voice technologies.
- AI and ML systems must identify and share high-probability suspected-UCC calling-line identifiers, but the regulation treats an AI flag as an investigation trigger rather than conclusive proof.
- Entities using A2P calls must pre-declare their use and calling-line ranges when the 60-day provision starts; undeclared A2P calls are to be treated as UCC under the amendment.
- A consumer appeal path and AI-corroborated three-complaint enforcement route are in the 90-day Annexure, while the up-to-Rs. 0.05-per-minute A2P charge is between access providers, not a stated consumer fee.
The rules are final, but their obligations do not start all at once
TRAI has notified the Telecom Commercial Communications Customer Preference (Third Amendment) Regulations, 2026, a final change to India’s commercial-communications framework that addresses AI-led suspected-spam detection, automated calls, complaint appeals and caller-ID applications. The Telecom Regulatory Authority of India lists the amendment with a release date of 18 September 2026. For mobile users, the immediate practical point is restraint: this is not evidence that every new control is already operating or that unwanted calls will stop at once.
The regulation sets its commencement by publication in the Official Gazette, not simply by its 18 September date. Most amendments are to start 30 days after that publication. A specified 60-day group includes declarations for Application-to-Person calls, the A2P termination-charge provision, a digital interface for virtual network operators, and some header and content-template measures. The Annexure puts the consumer appeal and the AI-corroborated complaint changes in a 90-day group. TRAI’s regulation index and the final text were checked on 20 September; neither source independently supplied a Gazette publication date. That is why calendar start dates should not be inferred from the notification date alone.
The policy process also matters. TRAI published a draft amendment for consultation on 13 March, received comments and counter-comments, and held an open-house discussion on 3 June. Its 18 September information note describes the final measure as an amendment to the 2018 Telecom Commercial Communications Customer Preference Regulations, commonly called TCCCPR. It is a final regulation after consultation, rather than a new consultation paper, but the staged legal timetable still determines when individual obligations apply.
AI flags are investigation triggers, not automatic proof of spam
The new Regulation 21A puts AI and machine-learning suspected-UCC detection into the regulations. A terminating access provider must use its AI/ML-based UCC_Detect system to identify calling-line identifiers with a high probability of being used for unsolicited commercial communication and share the relevant information with access providers through the distributed-ledger platform. If five or more CLIs associated with a sender are identified as suspected high-probability UCC CLIs within 10 days, originating providers must investigate and apply the regulation’s graded process.
That process is narrower than the phrase AI spam blocking can suggest. TRAI’s explanatory memorandum says AI-generated intelligence does not itself constitute conclusive evidence for punitive regulatory action because false positives cannot be ruled out. The system is intended to trigger further investigation, not act as a stand-alone conviction. Regulation 21A also requires an access provider to offer a flagged sender a way to make a representation with justification and to remove the flag if it finds the justification in order. This safeguard is material for legitimate businesses and call centres whose patterns may resemble automated outreach.
The technology is not entirely new. TRAI’s 27 February 2026 direction had already required institutional use of AI/ML-based UCC_Detect intelligence for sharing and action, after providers had deployed network-level detection and alert systems. The September amendment incorporates that framework into the regulations. TRAI’s own press note says major telecom service providers had already deployed AI/ML systems to detect suspected spam and alert customers. The change therefore formalises and extends an enforcement architecture; it should not be presented as the first day that telecom AI is looking for spam.
Robocalls and artificial-voice calls are expressly brought into the A2P definition
The amendment gives a formal definition to an Application-to-Person, or A2P, call. It is a voice call initiated by an application, software system or automated platform without direct human dialling and delivered to an individual telecom subscriber. The definition expressly includes autodialling, robocalls and pre-recorded or artificial-voice technologies. That means the TRAI robocall rules are concerned with how the call is initiated, not merely with whether a recipient finds its content unwelcome.
The definition has boundaries. The final regulation frames A2P around the absence of direct human dialling; it does not say that every call involving a software tool is automatically a robocall. In the explanatory material, TRAI records stakeholder concerns about service calls where a platform connects a live driver, delivery worker, doctor or support agent. The published rule remains the controlling text, and companies will need to assess their own calling arrangements against it once the relevant provision takes effect. This article does not determine the status of an individual business workflow.
The label also does not make every A2P call illegal. Commercial communication remains subject to the TCCCPR framework, including registration, consent and other obligations. What the amendment adds is a specific A2P category and a declaration regime. Readers should therefore distinguish an automated calling method from a finding that a particular call was unsolicited commercial communication. A caller can be subject to a regulatory duty without each call being proven spam, while an undeclared automated commercial call may face the amendment’s UCC treatment when the declaration provision starts.
A2P users must declare the automated use and the calling-line ranges
The rule’s operational requirement is pre-declaration. The amendment requires a sender using A2P calls to declare that use to the originating access provider, together with the details of the calling-line identifier ranges used for the calls. The originating provider must place that declaration and the relevant CLI-range information on the digital-ledger platform. TRAI’s press note says A2P calls made without the required prior declaration will be treated as UCC.
This obligation belongs to the 60-day category. It does not take effect simply because the regulation is dated 18 September. The legal clock begins from Official Gazette publication, and the official material checked before this article’s cutoff did not establish that date. Organisations using automated or artificial-voice systems should not rely on this news report as compliance instructions; their carrier arrangements, declared ranges and applicable regulatory notices remain the relevant operational records.
The declaration is intended to improve traceability. A range-level record can help the terminating provider identify A2P traffic and apply the separate provider-to-provider charging provision. It is not a consumer-facing caller-verification badge and does not guarantee that an automated call is useful, authorised under every law, or free from misuse. The regulation also says consent under TCCCPR is limited to the communication channel agreed between sender and customer and does not exempt a sender from obligations under the Digital Personal Data Protection Act, 2023.
The 5-paise A2P amount is a carrier charge, not a new consumer bill
Regulation 35A permits the terminating access provider to charge the originating access provider up to Rs. 0.05 per minute, calculated on a per-minute-pulse basis, for A2P calls. The maximum is five paise per minute, not five rupees, and the parties named in the provision are telecom providers. TRAI characterises the charge in its explanatory material as a deterrent rather than a cost-recovery mechanism or a retail price. The regulation does not state that a person receiving a call will be charged five paise.
The wording up to is important. It is a ceiling, and the rule adds that the charge cannot exceed the lowest tariff the terminating access provider offers to senders or telemarketers for whom it acts as originating provider. There are also stated exemptions: A2P calls using number resources from 140xx, 1600xx, 1601xx or another series designated by TRAI for commercial communications are excluded unless TRAI notifies otherwise. Calls made by or on the directions of TRAI, and calls made by an agency authorised by TRAI, are also excluded.
The charge likewise sits in the 60-day commencement group. It should not be reported as a new consumer levy, a fixed retail rate or an immediate price change. Whether, and how, providers set permitted interconnection charges after the legal start date is different from what Regulation 35A establishes. The confirmed fact today is the regulatory ceiling and the provider-to-provider structure, together with the specified exemptions.
Caller-ID apps must pass UCC reports into the telecom enforcement system
The amendment addresses call-management applications, a category that includes phone diallers and third-party apps. A call-management app cannot offer a user a mechanism to report a call as spam, junk or another term that implies unsolicited commercial communication unless the app sends that report to the digital-ledger platform maintained by access providers. The aim is to move relevant user reports into the telecom system that providers use for UCC detection and enforcement rather than leaving them only inside an app’s private reporting feature.
A separate restriction prevents blanket app-level tagging, blocking or filtering of calls from number series designated by TRAI or the Central Government for commercial or government communications. TRAI’s press note identifies 140xx for regulated promotional calls and 1600xx and 1601xx for service, transactional or government-related calls. The safeguard is designed to reduce the risk that legitimate regulated or government calls are broadly mislabelled. It is not a declaration that calls from those series are always safe, genuine or desired.
The consumer’s own device remains different from a blanket platform setting. The regulation expressly preserves a consumer’s freedom to filter or block any calls on that person’s device, provided that personal choice does not affect other recipients’ settings or experience. The app duties belong to the general 30-day commencement rule, so users should not assume that every caller-ID or dialler application has already changed its reporting path. The rule also gives a non-compliant application an opportunity to represent before TRAI initiates action under applicable laws.
Existing complaint routes remain available; the new appeal route comes later
Consumers do not need to wait for the amendment to report an unwanted commercial call or message. TRAI’s current UCC guidance says complaints or reports can be made to the telecom service provider by SMS to 1909, a call to 1909, IVRS, an approved mobile app, or an OTP-authenticated web portal. It distinguishes a complaint made within three days, which is considered valid for investigation, from a later report, which can help providers identify spam patterns. The provider is to issue a complaint ID so the consumer can track the submission and action information.
The amendment adds an appeal option to Regulation 23. A complainant will be able to appeal to the Appellate Authority under the Telecom Consumers Complaint Redressal Regulations, 2012 within 15 days of receiving information about the complaint’s resolution. If no redressal information is received within 15 days of lodging the complaint, the appeal can be made within the next 15 days after that period lapses. TRAI says the appeal can use the same modes available for a UCC complaint or report, including the TRAI DND App, provider apps or portals, and call or SMS to 1909.
That is a new route, not an instant remedy as of this check. The appeal amendment is listed in the 90-day Annexure, which is expressly measured from Gazette publication. The TRAI DND app already provides functions including crowdsourced data about offending messages and calls and updates about action taken on complaints, according to TRAI’s app page. Those existing functions should not be confused with a legally commenced new appeal process before the Annexure’s timing condition is met.
The three-complaint route requires AI corroboration
TRAI is lowering one complaint-based trigger, but not to three complaints in every case. Under the amended Regulation 25 route, action can be triggered when there are three or more complaints from unique recipients in the preceding 10 days and at least one CLI of the sender is also identified by an access provider’s AI/ML system as suspected high-probability UCC. The corroboration requirement is the evidence pair: consumer complaints provide one signal, while the provider’s AI/ML flag supplies the other.
The regulation retains a separate five-unique-recipient route. Where there are five or more complaints against a sender in 10 days, the provider’s response follows the regulation’s complaint-handling framework. Where complaints are below five and none of the sender’s CLIs has been flagged as suspected high-probability UCC in that period, the originating provider is to tell the terminating provider to inform the complainant of closure with reasons. These details are why a headline that simply says TRAI has made three complaints enough is incomplete.
The lower, AI-corroborated threshold is another 90-day Annexure change. It is intended to allow earlier action where two forms of evidence align, while maintaining a guard against action based on an isolated signal. It is not a guarantee that a report will produce disconnection, that every potentially unwanted call can be identified, or that providers will never make an error. The outcome depends on the facts of a complaint, the regulatory thresholds, investigation and the safeguards written into the rules.
What does not change immediately and what to watch next
The confirmed news is a final TRAI amendment dated 18 September, not a completed nationwide enforcement event. It does not establish automatic AI blocking, a universal safe-number list, a new fee to a call recipient or a guaranteed end to spam calls. It also does not convert the notification date into the start date of the 30-, 60- or 90-day groups. A consumer can continue to use existing UCC complaint and preference tools, while providers, senders and applications have to follow the legal commencement schedule for the new duties.
The most useful next official signal is the Gazette publication record that fixes the statutory reference point for the phased provisions. Subsequent TRAI notices, provider implementation information and app updates can then show how the declaration, reporting and appeal functions are being put into practice. For the designated 140xx, 1600xx and 1601xx series, the rule changes the limits on blanket app treatment but does not replace personal judgement: a user may still block a call on that user’s own device.
For now, the plain-English conclusion is limited. TRAI has created a more technology-led anti-spam framework that connects provider AI signals, complaints, A2P traceability and app reports. The regulation itself limits the promise: AI needs investigation, key measures are phased, and individual consumers keep their existing reporting options. That distinction is central to understanding the TRAI anti-spam rules without overstating either their immediate reach or their likely effect.
Reader guide
Article questions, answered
Short answers to common reader questions based on the reporting above.
When do TRAI's new anti-spam rules start?
The Third Amendment Regulations state that most provisions begin 30 days after publication in the Official Gazette. Specified A2P declaration, VNO-interface, header/template and A2P-charge provisions begin after 60 days, while the consumer-appeal and AI-corroborated complaint changes in the Annexure begin after 90 days. As of the 20 September 2026 IST source check, the official materials checked did not independently establish a Gazette publication date, so those intervals should not be converted into calendar dates.
Do AI checks let TRAI automatically disconnect a number as spam?
No. Regulation 21A requires access providers to identify and share high-probability suspected-UCC calling-line identifiers, and a five-CLI pattern within 10 days can trigger investigation and graded action. The regulation's explanatory material says AI-generated intelligence is not conclusive evidence by itself because false positives are possible. Flagged senders must have a way to make a representation, and the provider can unflag a CLI if the justification is accepted.
How can a consumer complain about an unwanted commercial call or message?
TRAI says a consumer may complain or report UCC to the telecom service provider by SMS to 1909, calling 1909, IVRS, an approved mobile app or an OTP-authenticated web portal. A complaint made within three days is treated as a valid complaint for investigation; a later submission is treated as a report that can help identify trends. The new appeal route, once its 90-day Gazette-linked commencement applies, will cover complaint-resolution information or a failure to receive it within the stated period.
Is the up-to-5-paise A2P charge a new fee for people receiving calls?
No. Regulation 35A permits a terminating access provider to charge the originating access provider up to Rs. 0.05 per minute for A2P calls. It is a provider-to-provider ceiling, not a stated retail charge to the person receiving a call. The regulation also lists exemptions, including calls using designated 140xx, 1600xx and 1601xx series unless TRAI notifies otherwise, and calls made by or for TRAI-authorised agencies.
Sources and further reading
These references support the factual context used in this article. Links open the original publisher.
- The Telecom Commercial Communications Customer Preference (Third Amendment) Regulations, 2026 (05 of 2026)Telecom Regulatory Authority of India · accessed 20 September 2026
- RegulationsTelecom Regulatory Authority of India · accessed 20 September 2026
- TRAI Strengthens Framework for Curbing Unsolicited Commercial Communications through Technology-Driven Enforcement and Enhanced Consumer ProtectionTelecom Regulatory Authority of India · accessed 20 September 2026
- Consultation on Draft Telecom Commercial Communications Customer Preference (Third Amendment) Regulations, 2026Telecom Regulatory Authority of India · accessed 20 September 2026
- Open House Discussion: Rescheduled OHD on Consultation on Draft Telecom Commercial Communications Customer Preference (Third Amendment) Regulations, 2026Telecom Regulatory Authority of India · accessed 20 September 2026
- Direction regarding institutionalization of AI/ML-based UCC_Detect intelligence for inter-operator sharing and regulatory action against UCC sendersTelecom Regulatory Authority of India · accessed 20 September 2026
- Complain or report against UCCTelecom Regulatory Authority of India · accessed 20 September 2026
- Portals and Apps: TRAI DND (Do Not Disturb)Telecom Regulatory Authority of India · accessed 20 September 2026
- Direction regarding enhancing user-friendliness of registration of UCC complaints, preferences and consents through mobile apps and web portals of access providersTelecom Regulatory Authority of India · accessed 20 September 2026
- Regulation and Amendment: Telecom Commercial Communications Customer Preference RegulationsTelecom Regulatory Authority of India · accessed 20 September 2026
