Do I need to register on DLT before making AI outbound calls in India?

Yes — for any commercial communication, DLT registration of the Principal Entity, the calling header, the call template, and the telemarketer is required before the call can legitimately be placed. Transactional service calls operate under a lighter framework but still benefit from PE registration and from being routed through registered headers. The cost of registration is modest (typically Rs 5,900-7,500 for the PE plus minor amounts per template) and the timeline is 2-5 business days for PE plus 24-72 hours for templates. Most reputable AI calling platforms register as your telemarketer as part of standard onboarding.

What is the difference between 1600 and 140x number series for AI calling?

The 1600-series is reserved for transactional and service-related communications — order updates, OTPs, EMI reminders for existing loans, appointment confirmations, delivery notifications. These calls relate to a transaction the customer initiated and are exempt from DND scrubbing. The 140x-series is for commercial and promotional communications — marketing, upsell, cross-sell, abandoned cart recovery, win-back. Promotional calls require DND scrubbing, DLT template registration, and the 9am-9pm time-window restriction. Routing the wrong call type through the wrong number series is itself a compliance violation regardless of the call's content.

How often must I scrub my calling list against the NDND registry?

Before every campaign. The NDND database refreshes daily, so a list scrubbed two days ago will already be stale. Scrubbing should happen as the immediately preceding step before campaign launch, not as a one-time setup task. Scrubbing must also be at category level (filtering both all-DND consumers and those who have opted out of your specific industry category) rather than only against full DND. Scrubbing costs run Rs 0.01-0.05 per number — the cost of skipping the step is Rs 25,000 per upheld complaint, so the economics are not subtle.

Are abandoned cart recovery calls considered promotional under TRAI?

Yes. Abandoned cart recovery calls are classified as promotional commercial communications because the customer did not complete a transaction — they added items to a cart but did not buy. The call's purpose is to recover a sale that did not happen, which is marketing-coded under TCCP. Cart recovery calls require DND scrubbing, DLT template registration, 140x number series routing, the 9am-9pm time window, and explicit promotional consent. Many D2C brands incorrectly classify cart recovery as transactional and run unscrubbed campaigns; this is one of the most common TRAI compliance failures in the e-commerce sector.

What is the penalty for calling a DND number with an AI voice bot?

Up to Rs 25,000 per upheld complaint. The penalty is levied on the Principal Entity, not the calling platform. For repeat or systemic violations, the consequences escalate to suspension or blacklisting of the PE registration, which prevents the entity from placing any commercial communication on Indian telecom networks — a business continuity event for any calling-dependent business. The penalty is the same whether the call was placed by a human telecaller or an AI voice agent; TRAI does not distinguish between human-mediated and AI-mediated commercial communications for purposes of the penalty regime.

Does TRAI require AI calling platforms to disclose they are automated?

Not yet by formal regulation, but the trajectory is unambiguous. TRAI has an active consultation on AI-generated and synthetic-voice communications, and the likely outcome includes a mandatory AI-disclosure requirement at the start of every commercial call. The EU AI Act already requires this disclosure, and global regulatory direction is converging. Build the disclosure into your AI scripts now — a simple opening line acknowledging the AI nature of the call. Customer acceptance rates for transparent AI disclosure consistently run in the 82-88% range.

How do TRAI and DPDP regulations interact for AI outbound calling?

TRAI's TCCCPR framework governs the act of communication — registration, scrubbing, number series, templates, time windows. DPDP governs the underlying data processing — consent, purpose limitation, data residency, opt-out propagation, grievance redressal. The two frameworks interact in several ways: TRAI's distinction between transactional and promotional calls maps closely to DPDP's distinction between Section 7 legitimate use and Section 6 explicit consent; DPDP's opt-out cascade requirements reinforce TRAI's suppression list obligations; both frameworks require the consent to be specific to the purpose. A compliant AI calling programme has to satisfy both frameworks simultaneously.

TRAI DND Compliance for AI Outbound Calling India 2026 | Caller Digital

The ₹25,000 figure sounds abstract until you do the multiplication. A 10,000-call campaign placed to an unscrubbed list, with a one percent overlap with the National Do Not Disturb registry, produces 100 potentially non-compliant calls. At ₹25,000 per upheld complaint, that is a ₹25 lakh exposure on a single campaign. Run that campaign weekly for a year, and the math gets the kind of attention that puts compliance on a CFO's quarterly board update.

This is the unglamorous, operationally critical compliance question that most AI calling deployments in India have not solved. Every voice AI platform will help you place a call. Very few will tell you, before the call goes out, whether you are legally allowed to place it. The gap between "the platform can dial this number" and "you are authorised to dial this number" is where TRAI compliance lives.

This article is the field manual. Not a treatise. Not a legal opinion — for that, talk to your counsel. A practical, structured walk through what TRAI's regulations actually require, where AI calling deployments quietly violate them, and what a compliant calling stack looks like in 2026.

TRAI's Regulatory Framework — The Picture That Matters

The Telecom Regulatory Authority of India regulates commercial communications under the Telecom Commercial Communications Customer Preference Regulations, commonly abbreviated TCCCPR. The current regulation in force is TCCCPR 2018, which replaced the earlier 2010 framework and added the digital ledger backbone that runs on a blockchain-based DLT (Distributed Ledger Technology) platform.

The framework rests on six pillars that you need to understand before placing a single outbound call.

The DND registry, formally the National Do Not Disturb registry, is the central database of consumer preferences. Indian customers can register their phone numbers as DND through their telecom operator, the DoT's Sanchar Saathi portal, or the 1909 number. Once registered, those numbers cannot receive promotional commercial communications. The registry refreshes daily; a number that was not DND yesterday might be DND today.

Category-level preferences are the often-missed layer beneath all-DND. Indian consumers can opt out of specific categories — banking and insurance, real estate, education, health, consumer goods, communication and broadcasting — rather than blocking all promotional calls. A number that is "not on full DND" might still have category preferences excluding your specific industry. Your scrubbing logic must respect category-level preferences, not just full DND.

The DLT (Distributed Ledger Technology) platform is TRAI's blockchain-based infrastructure for registering principal entities, telemarketers, headers and message templates. Six telecom operators host DLT platforms — Airtel, Jio, Vi, BSNL, Tata, and Videocon — and registrations on any one mirror to all. DLT is where every legitimate commercial communication in India is supposed to be pre-registered before transmission.

Principal Entity (PE) and Telemarketer registration is the formal recognition of who is making the call. The PE is the brand whose communication is being sent — your D2C brand, your NBFC, your hospital. The telemarketer is the entity that places the calls on the PE's behalf — typically your AI calling platform. PE and telemarketer must both be registered, and the linkage between them must be active for any commercial call to be legitimate.

Number series classification distinguishes service from commercial calls at the dialling layer. The 1600-series numbers are reserved for transactional/service calls — these are the numbers your AI should be using for COD confirmation, delivery updates, OTPs, EMI reminders. The 140x-series numbers (140, 141, 142, etc.) are for commercial/promotional calls. Routing the wrong call type through the wrong number series is a violation regardless of content.

Time-window restrictions limit promotional calls to the 9am-9pm window. Transactional calls are exempt from this restriction, but their classification has to be defensible — calling a customer at 10pm with a "delivery update" that is really a marketing offer fails on classification grounds.

The Most Important Distinction in TRAI Compliance: Transactional vs Promotional

If you understand only one thing about TRAI's framework, understand this. The distinction between transactional and promotional calls determines almost every downstream compliance question — number series, DND scrubbing, time-window restrictions, DLT template registration requirements. Get the classification right and most of your compliance work is done. Get it wrong and the rest of the framework collapses around you.

A transactional call directly relates to a transaction the customer initiated. The defining characteristic is causation: the customer's prior action triggered the call, and the call serves that prior action.

Examples that are unambiguously transactional:

A delivery confirmation call to a customer who placed a COD order
An OTP for a payment the customer is currently making
An EMI due reminder on an existing loan the customer signed
An appointment confirmation for a service the patient booked
A shipment status update for an order in transit
A balance alert for a transaction the customer just performed
A flight delay notification for a ticket the customer holds

A promotional call is marketing, advertising, or upsell — content the customer has not transactionally initiated.

Examples that are unambiguously promotional:

A new product launch announcement
An offer for a 20% discount on a different product category
An abandoned cart recovery call for a cart the customer never converted
A win-back call to a lapsed customer
An upsell pitch for a higher-tier subscription
A cross-sell call for an unrelated product
A real estate developer's "new launch" call to a database of past leads

The grey zone is where most violations happen. Three patterns to flag specifically.

The post-purchase upsell embedded in a confirmation call. A logistics partner calls to confirm a delivery time, and at the end of the call says "and would you like to upgrade to our premium subscription for ₹99/month?" The promotional content converts the entire call's classification to promotional. The campaign was likely filed as transactional and dialled without DND scrubbing. The violation is sitting in the call recording, queryable by anyone who files a complaint.

The "important update" that is actually a marketing communication. A bank calls a customer about an "important update to your account" that turns out to be a credit card offer. The framing is transactional; the content is promotional. The classification of the call is determined by the actual content delivered, not the framing in the opening line.

The reactivation call to a lapsed customer. The customer transacted once, two years ago. The brand calls to "check in" and offer a personalised discount. The original transaction has long since concluded. This is a promotional re-engagement call requiring promotional consent and DND scrubbing, regardless of the historical relationship.

When in doubt, classify as promotional and apply the stricter compliance posture. The cost of unnecessary scrubbing is negligible. The cost of misclassification is ₹25,000 per upheld complaint plus reputational damage and, for repeat offenders, telecom blacklisting.

DLT Registration: Step by Step

Here is what the registration process actually looks like, in operational sequence.

Step one: register as a Principal Entity. The PE registration is filed on any one of the six telecom-operator DLT platforms. Documentation required: certificate of incorporation, GST registration, PAN of the entity, details of authorised signatories, and a one-time registration fee (typically ₹5,900 to ₹7,500 depending on platform). Verification timeline: 2-5 business days. Once verified, your PE registration mirrors automatically across all six DLT platforms.

Step two: register your headers. Headers are the entity identifier customers see at the call origination point — for SMS this is the sender ID (e.g., "ICICIB"); for voice this is your registered number series. Headers must be linked to the PE registration. Each header takes 24-48 hours to verify after submission. Multiple headers can be registered for a single PE.

Step three: register your call templates. This is where most teams lose time. Every promotional call script must be pre-registered as a DLT template before the call goes live. The template specifies the fixed content of the call and identifies any variable fields (customer name, amount, date, product) that get filled in at call time. Templates are submitted with a category tag (banking, real estate, education, etc.) and an associated header. Approval takes 24-72 hours; rejected templates need rework and resubmission.

Step four: link your telemarketer. Your AI calling platform must be registered on DLT as a telemarketer and explicitly linked to your PE registration. Without this linkage, calls placed by the platform on your behalf are technically unauthorised. Caller Digital handles this linkage as part of standard onboarding, but if you are evaluating other platforms, ask specifically whether telemarketer registration is included or whether you have to coordinate it separately.

Step five: link templates to campaigns. Each campaign that runs through the calling platform must reference an approved template. The campaign cannot launch if its associated template is not in approved status. Campaigns running on unregistered or rejected templates are unauthorised at the dial layer — most modern platforms will refuse to launch them, but if your platform allows it, the violation is still on you as the PE.

Common registration mistakes:

Templates registered with placeholder variables that don't match what the call actually says. The DLT framework requires fidelity between registered template and live content. If your template says "Dear {customer_name}, your order #{order_id} has been confirmed" and the live call says something materially different, you are non-compliant against your own template.

Failure to link the calling platform as the registered telemarketer. Common in shadow IT setups where marketing teams set up campaigns without informing compliance.

Using one template for two different campaigns with different content. Each substantively different script needs its own template.

"Test campaigns" on unregistered templates. There is no test exemption. A test campaign that places real calls to real customers is subject to the same registration requirements as production.

NDND Scrubbing: The Operational Discipline

Scrubbing is the operational act of filtering your dial list against the National Do Not Disturb registry before launching a campaign. The discipline that separates compliant operations from non-compliant ones is not whether you scrub, but how often.

Scrub before every campaign. Not once at platform setup. Not weekly. Before every campaign. The DND database refreshes daily; a number that was not DND yesterday might be DND today. A campaign that scrubbed its list two days ago, then dials today, is dialling against stale data.

Scrub at the right granularity. All-DND scrubbing removes consumers who have opted out of all promotional calls. Category-level scrubbing additionally removes consumers who have opted out of your specific category (e.g., all banking and insurance promotional calls). Without category-level scrubbing, you can technically be calling consumers who are not on full DND but who have specifically blocked your industry. Modern scrubbing services support both layers; the laggard practice of scrubbing only against full DND is a violation waiting to be filed.

Maintain your internal suppression list. Beyond the TRAI registry, you have your own do-not-call list — customers who have opted out of communications from you specifically. This list must be independent of TRAI scrubbing and applied as an additional filter. A customer who is not DND-registered but who told your AI agent "don't call me again" three months ago must still be excluded from this campaign. Internal suppression list management is where many programmes fail; the operational propagation from "customer said don't call" to "next campaign excludes this number" is fragile.

Honour the opt-out across channels. A customer who opts out of voice calls is presumed in current regulatory posture to have opted out of voice. They have not necessarily opted out of WhatsApp or email — those are separate consent layers. But the trajectory of regulation, both DPDP and TRAI, points toward harmonised opt-out across channels. A customer-friendly compliance posture treats voice opt-out as a signal to also pause SMS and WhatsApp marketing pending confirmation.

Scrubbing economics. Per-number scrubbing cost runs ₹0.01 to ₹0.05 depending on volume and provider. On a 100,000-number list, that is ₹1,000 to ₹5,000. The cost of skipping the scrub: ₹25,000 per upheld complaint. The math is not subtle.

The Operational Compliance Checklist

Twelve questions that determine whether your AI calling programme is in compliant state right now. Walk through these with your ops lead and your compliance counsel.

Is your outbound number series correct — 1600 for service/transactional, 140x for promotional/commercial?
Are you registered as a Principal Entity on DLT? Can you produce the PE registration ID on demand?
Is your AI calling platform registered as your telemarketer on DLT, with active linkage to your PE?
Is every promotional call script registered as an approved DLT template, with the variable fields matching what the call actually says?
Is NDND scrubbing happening before every campaign — not just at platform setup, not just weekly, but every campaign?
Is scrubbing happening at category level, not just against full-DND?
Are promotional calls restricted to the 9am-9pm window, with no exceptions for "test runs" or "urgent campaigns"?
Is your AI disclosing it is an automated call within the first 30 seconds?
Is there a keypress or spoken opt-out option in every promotional call, with the opt-out reliably detected by the AI?
Are opt-outs propagating to your CRM and suppression list within 24 hours?
Are call recordings retained for at least 90 days, with the ability to retrieve any specific call on inspection request?
Do you have a designated nodal officer for UCC complaints, with their contact details published and easily findable?

Score nine or more, your programme is in better shape than most. Score five or fewer, you have remediation work, and the work is best done before a complaint forces it.

What Happens When You Get a Complaint

The complaint pathway works like this in 2026.

A consumer who believes they have received an unauthorised commercial call files a complaint via one of three routes: the DoT Sanchar Saathi portal, the 1909 complaint number, or directly with their telecom operator. The complaint is logged with the originating number, the time of the call, the consumer's number, and a brief description.

The telecom operator forwards the complaint to the Principal Entity associated with the originating header, typically within 3 days. The PE has 15 days to respond and resolve.

Resolution requires producing evidence of authorisation: the registration of the PE, the registration of the calling number, the registration and approval status of the call template, the consent record (for promotional calls), and proof of NDND scrubbing for the campaign. Most non-compliant programmes fail at the consent and scrubbing layers — the records simply don't exist.

Unresolved or upheld complaints carry a financial penalty of up to ₹25,000 per complaint. The penalty is levied on the PE, not the telemarketer.

For repeat offenders, the consequences escalate beyond fines. The DLT platforms can suspend the PE registration, which effectively prevents the entity from sending any commercial communication on Indian telecom networks. For a calling-dependent business, this is a business continuity event.

A realistic worst case for a non-compliant programme: a 50,000-number campaign placed to a list with 1% DND overlap and no category-level scrubbing. Of the 500 potentially non-compliant calls, perhaps 50 generate complaints (10% complaint rate is on the high end but realistic for aggressive promotional content). All 50 are upheld due to absent scrubbing records. Total exposure: ₹12.5 lakh on a single campaign. Plus reputational damage, plus the operational cost of responding to 50 complaints in 15 days, plus the increased regulatory scrutiny that follows.

The economics of non-compliance versus the economics of compliance is not a close question.

The Special Cases

A few scenarios where the framework has nuance worth understanding.

Inbound calls and missed calls. When a customer initiates the call to you, TRAI's outbound regulations don't apply — you are not making a commercial communication, you are answering one. AI inbound and missed-call handling is therefore in a less regulated zone, though DPDP and sectoral overlays still apply to data processing and consent. Missed-call callback automation operates in this zone with appropriate care.

Existing customer service calls. A bank calling its existing customer about that customer's loan account is unambiguously transactional, regardless of the call's specific topic. Service relationship is the operative principle. The grey zone arises when the bank tries to cross-sell a different product on the same call.

Welcome and onboarding calls. A new customer who just signed up for a service is in an active service relationship. Welcome and onboarding calls for that service are transactional. Welcome calls that pivot to upselling additional products convert to promotional.

B2B calls. Calls between businesses, where the called number is registered to a business and not an individual consumer, are outside the scope of TCCCPR's consumer-protective framework. This does not exempt B2B AI calling from DPDP; it just changes the TRAI calculus. Most B2B calling programmes still benefit from observing TRAI norms voluntarily — number series, time windows, AI disclosure — for reputation and customer experience reasons even where strict compliance is not required.

International calls into India. Calls originating from outside India to Indian consumers are increasingly subject to TRAI scrutiny under the new international long-distance framework. Global voice AI platforms placing calls into India should be aware that the "we're not based in India" defence is becoming harder to sustain.

The Direction of Regulatory Travel

TRAI's posture in 2026 is hardening, not softening. Three trends to plan around.

AI-specific calling regulations are coming. TRAI's consultation paper on AI-generated and synthetic-voice communications has been in process for over a year. The likely outcomes: a mandatory AI-disclosure requirement at the start of every commercial call, additional consent requirements for the use of synthetic voices, and recordkeeping obligations specific to AI-generated content. Build the disclosure into your scripts now — you are very likely to be required to in 2026 or early 2027 anyway.

Consent harmonisation across DPDP and TCCP. The two frameworks were drafted with somewhat different consent architectures. The operational reality of running an AI calling programme requires reconciling them — and the regulatory direction is toward harmonisation, with consent meaning roughly the same thing under DPDP as under TCCP. Programmes that build on the harmonised foundation now will have less rework when the formal alignment arrives.

Caller ID transparency. TRAI has signalled increasing concern about spoofed caller IDs and number-spoofing scams. The trajectory is toward verified caller display — where the called consumer sees not just a number but a verified entity name. For legitimate AI calling programmes, this is a positive development; spoofers and scammers face higher friction. Operationally, expect to need additional verification at the PE-registration layer.

Cross-channel orchestration regulation. The regulator's attention is expanding from voice to the cross-channel orchestration of voice plus SMS plus WhatsApp plus email. A customer who opts out of voice may not have opted out of WhatsApp, but the regulatory question of whether one opt-out should extend to others is being asked. The customer-friendly answer is yes; the legal answer is evolving.

How Caller Digital Handles TRAI for You

Brief, because this is the section where vendor-published material gets self-serving fast.

Caller Digital's platform handles the TRAI compliance layer as an architectural feature, not as a customer responsibility. NDND scrubbing runs before every campaign automatically, with category-level filtering enabled by default. DLT template management is integrated into the campaign creation workflow — you cannot launch a campaign without an approved template linked. The platform classifies each campaign as transactional or promotional and routes to the correct number series automatically. Opt-outs detected mid-call propagate to the CRM and suppression list within four hours. Call recordings are retained per TRAI's 90-day minimum on Indian infrastructure, with retrieval workflows for inspection requests. Telemarketer registration is included in onboarding; we register as your telemarketer linked to your PE during the first week of deployment.

For deeper context on how TRAI interacts with the DPDP Act for AI calling, and on the sector-specific overlays for insurance under IRDAI, the linked guides are companion reading. For the broader voice AI India 2026 picture, the pillar guide pulls the threads together.

The closing observation: TRAI compliance is not a feature you bolt on. It is an operating discipline. The platforms that build it in by default save you the work; the platforms that don't push the work onto your team. For Indian businesses running production-scale AI calling, the cost of getting this right is small. The cost of getting it wrong is the kind that ends up on a board agenda.

TRAI DND, DLT and TCCP Compliance for AI Outbound Calling: The 2026 Field Manual for India