A D2C brand in Indore builds an AI voice agent to call a purchased list of leads and pitch a new product, and it works, the agent sounds natural, books a handful of meetings in its first two days, and the founder is ready to scale it to a thousand calls a day. By the end of week one, the business has three TRAI complaints filed against it and a notice threatening a fine. Nobody on the team thought to check whether an AI-generated voice calling a number on the National DND registry counted as the same violation a human telemarketer would commit. It does, and in 2026 the rules got sharper, not softer.
Is calling with an AI voice agent actually legal in India?
Yes, but only inside the same framework that governs any commercial telemarketing call. In India, TRAI's DND rules now apply to AI-driven outbound calls under unsolicited commercial traffic guidelines, and AI-generated voices are officially classified as "artificial" voices under existing robocall regulations, which means every AI calling platform must follow the same rules as traditional automated dialing systems. There is no separate, lighter-touch category for a call just because a language model generated the script.
That surprises founders who assume AI tooling operates in a regulatory gray zone simply because it's new. It doesn't. The call is still a commercial communication to an Indian telecom subscriber, and TRAI's Telecom Commercial Communications Customer Preference Regulations (TCCCPR) govern it regardless of what's on the other end of the line generating the words.
What is the National DND registry, and what happens if an AI agent calls a number on it?
The National Customer Preference Register (NCPR, commonly called DND) is the list of subscribers who've opted out of promotional calls and messages. Calling a DND-registered number for promotional purposes triggers an immediate complaint, and businesses relying on third-party or purchased lead lists carry the highest risk since those numbers may have opted into DND after the list was sold, meaning a list that was clean six months ago can be full of violations today.
The practical requirement is a scrub-before-dial step: every number in a calling list needs to be checked against the current NCPR immediately before each outbound campaign runs, not once at list purchase time, because DND status changes continuously and yesterday's clean list is not a guarantee for today's run.
What consent actually covers an AI agent calling a lead who isn't on DND?
- Explicit consent is valid for only 7 days for commercial transactions under the 2025 TCCCPR amendments, so a lead who filled out a form two weeks ago is no longer a valid basis for an outbound call today, even if they were never on DND.
- Inferred consent, from an existing customer relationship, is valid only for the duration of that contractual relationship, not indefinitely after the relationship ends.
- The Digital Consent Registry is a centralized, interoperable repository of consents maintained by telecom service providers, meaning consent increasingly needs to be recorded in a system telecom providers can actually verify, not just logged in a company's own CRM.
- Promotional calls must run through 140-series numbers, and BFSI (banking, financial services, insurance) service calls must use 1600-series numbers as of the January 2026 deadline, so an AI agent dialing out from a normal business line for promotional purposes is itself a compliance gap before a single word is spoken.
What are the actual penalties, and how aggressively is this being enforced?
Non-compliance carries financial penalties of up to Rs 10 lakh and immediate service suspension, and enforcement has gotten faster, not slower: the February 2025 amendment to the TCCCPR framework gave TRAI the ability to act directly against violating Principal Entities without routing enforcement through Access Providers, cutting the lag between a filed complaint and an actual consequence.
TRAI also prohibits all commercial calls before 9:00 AM or after 9:00 PM local time, every day of the week including weekends and public holidays, a rule an AI agent's calling schedule needs to enforce programmatically, not rely on someone remembering to configure correctly.
What does a compliant AI voice calling setup actually look like?
- An automated NCPR scrub runs against the full call list immediately before every campaign, not at list-acquisition time.
- Consent timestamps are tracked per lead, and any lead past the 7-day explicit consent window is automatically excluded from the queue rather than called on stale consent.
- The agent dials from a registered 140-series number for promotional traffic, or 1600-series for BFSI service calls, never from a standard business line.
- Calling hours are hard-coded into the scheduling logic (9 AM-9 PM local time), so a queued call can't fire outside the legal window even if a campaign is left running overnight.
- Every call, consent basis, and DND check is logged, so a TRAI complaint can be answered with a specific record instead of a shrug.
How AIBOOTSTRAPPER helps
AIBOOTSTRAPPER's AI automation team builds outbound AI voice agents with NCPR scrubbing, consent-window enforcement and calling-hour restrictions wired into the workflow itself, the same discipline covered for inbound voice agent architecture, applied to the outbound side where TRAI compliance is non-negotiable, not optional.
If you're planning an AI outbound calling campaign in India, or already running one without a documented consent and DND process, book a call and we'll review the compliance gap before TRAI does.
Want this done for you?
Book a free strategy call and we'll show you how to build and market your business with AI.