On 18 September 2026, the Telecom Regulatory Authority of India (TRAI) amended its spam framework through the Telecom Commercial Communications Customer Preference (Third Amendment) Regulations, 2026. Call-management apps such as Truecaller may no longer tag, filter or block calls from the 140xx, 1600xx and 1601xx series. They must also send user spam reports to the telecom industry's DLT platform rather than keeping them in-app. Automated and AI-voice calls are reclassified as application-to-person (A2P) traffic with a termination charge of up to five paise a minute.
Some of this is sensible. Some of it quietly moves control from users to carriers.
The case for the rules
The strongest argument for TRAI is that the number series in question are accountable by design. In its press release, TRAI says the 140xx, 1600xx and 1601xx series are designated for commercial communications, and that keeping them from being flagged as suspected spam ensures "legitimate commercial communications and government communications" are not "inadvertently missed." Anyone who has missed a bank fraud alert or a delivery OTP because a crowd-sourced label said "spam" will see the point. Crowd-sourced labels are also error-prone: a label applied to one caller can stick to a legitimate business for months.
The amendment also contains genuinely useful measures. A new Regulation 21A obliges telecom providers to identify sender numbers with a high probability of being used for spam and share that intelligence between operators. The complaint threshold for action falls from five unique complaints in ten days to three when the sender's number is also flagged by the operator's AI/ML system. Consumers get a right to appeal a wrongly closed complaint within 15 days, through the TRAI DND app, operator portals, or by calling or texting 1909. These are proportionate steps, and the consultation was orderly: a draft on 13 March, comments to 19 April, counter-comments to 4 May, and an open house on 3 June, per the press release.
Where the design goes wrong
The first problem is the blanket bar. MediaNama reports that 140-series calls can be blocked only where customers have opted out through the DND registry, and that 1600-series calls cannot be blocked by apps at all. That treats a number prefix as proof of good conduct. A prefix says how a sender was registered, not how they behave. A user who receives repeated unwanted calls from a 140 number now has fewer tools than before, and must route the complaint through the same operator system that critics say let the problem grow. Users, not regulators, are best placed to decide which registered callers they want to hear from.
The second problem is the one-way data flow. According to TechCrunch, Truecaller called the reporting requirement anti-competitive and "one-way," saying it hands commercially valuable data to operators. The company also said it has been compliant "since late last year" and that spam has "skyrocketed" because of what it called a free pass to spammers. Those are the company's claims and are not independently established, and Truecaller has an obvious commercial interest. Still, the structural point stands. India is Truecaller's largest market, with over 350 million of its 500-plus million monthly users, according to TechCrunch. A regulation that makes one class of intermediary a compulsory data supplier to another, without reciprocal access, deserves a competition analysis that the press release does not contain.
The third problem is incentives. Registered commercial series are exempt from A2P termination charges and shielded from app-level blocking. A bad actor who can obtain or misuse such a series inherits both protections. TRAI's answer is enforcement through the operators' AI systems and the complaint threshold. That may work, but the press release publishes no evidence that it works well enough to justify removing the app-level check. TRAI should publish the false-positive and false-negative rates of the operator systems it now leans on.
Pricing automated calls: right idea, blunt definition
Charging for A2P traffic is the most defensible new lever. The press release defines A2P calls as voice calls initiated by an application, software system or automated platform without direct human dialing, including autodialing, robo-calls and pre-recorded or artificial voice technologies. Senders must pre-declare such use and the numbers involved, or the calls are treated as unsolicited communications. Terminating operators may charge originating operators up to 5 paise a minute, which MediaNama says mirrors commercial SMS rates. A small per-minute cost changes the economics of mass robo-dialling.
The risk is collateral damage. The definition is technology-neutral in wording, so it also reaches an AI voice agent that confirms appointments, a clinic's reminder system, or an accessibility tool. TRAI's fix is pre-declaration, which is reasonable for businesses but adds friction for small developers. Regulators should confirm in operating guidance that consented, low-volume, transactional automation is not swept in as spam by default.
What to watch
MediaNama reports a phased rollout: 30 days for AI detection, A2P pre-declaration and the appeal mechanism; 60 days for A2P termination charges; and 90 days for the lower complaint threshold. Three tests will show whether the package worked:
- Whether spam complaints fall after app-level filters come off registered series.
- Whether operators publish accuracy data for their AI flagging.
- Whether app developers get a reciprocal data-access route or a written competition assessment.
India's spam problem is real, and operator-side enforcement is a legitimate response. But the better design keeps user-controlled filtering alongside it. TRAI can price robo-calls and hold operators accountable without switching off the one tool that lets a person decide who reaches their phone.