India's Department of Telecommunications (DoT) told the Rajya Sabha on July 23, 2026 that its in-house AI and big-data tool, ASTR, has flagged and disconnected 8.8 million (88 lakh) mobile connections that failed mandatory re-verification. A parallel channel, the crowd-sourced Chakshu fraud-reporting facility on the Sanchar Saathi portal, accounted for another 5.09 million (50.9 lakh) disconnections as of July 15, 2026. Union Minister of State for Communications Dr. Chandra Sekhar Pemmasani gave the figures in a written reply outlining the government's anti-cyber-fraud measures (Medianama; Free Press Journal). Combined, that's nearly 14 million SIM connections severed by an algorithm-and-crowdsourcing pipeline — one of the largest automated disconnection drives attempted by any telecom regulator globally.
What ASTR Actually Does
ASTR — Artificial Intelligence and big-data analytics for the enforcement of Telecom Subscriber verification norms — isn't new; DoT built it alongside Sanchar Saathi's 2023 launch. It scans subscriber and usage patterns across all licensed telecom networks to flag connections that look forged, duplicated, or non-compliant with KYC norms, then refers them to the operator for re-verification. Only connections that fail that second check get disconnected. Chakshu works differently: it lets citizens report a specific fraudulent call, SMS, or WhatsApp message, and DoT aggregates those individual reports into patterns before ordering cuts, rather than acting on any single complaint (Medianama). The official Sanchar Saathi portal describes this as part of a broader stack that also includes CEIR, which blocks stolen handsets, and TAFCOP, which lets subscribers audit every SIM issued in their name (sancharsaathi.gov.in).
The Case for the Crackdown
The strongest argument for ASTR is straightforward: SIM-based fraud — mule numbers used for OTP theft, phishing, "digital arrest" scams, and spoofed caller IDs — has become one of India's most common vectors for financial crime, and it scales through exactly the kind of bulk-registered, loosely verified connections ASTR is built to catch. A single fraud ring can burn through hundreds of SIMs registered on forged or borrowed documents; manual enforcement, operator by operator, cannot keep pace with that volume. DoT has also linked ASTR's outputs into a Digital Intelligence Platform shared with the Home Ministry's cybercrime unit (I4C), SEBI, and CERT-In, which is a sensible design — it means a number flagged for telecom fraud can also inform financial-fraud and market-manipulation enforcement rather than sitting in a silo (Free Press Journal).
Where the Concerns Lie
The problem is not that DoT is using AI to triage fraud at scale — that's a legitimate, arguably necessary, use of automation. The problem is that DoT has published the output (8.8 million disconnections) without publishing the inputs: no disclosed criteria for what usage pattern trips a flag, no error-rate disclosure, and no public accounting of how many "re-verifications" genuine subscribers failed simply because they lacked easy access to updated KYC documents — a real risk given how much of ASTR's reach extends into rural and lower-income subscriber bases that are more likely to be under-documented, not more likely to be fraudulent. A disconnection drive of this size, run opaquely, will misclassify some non-trivial number of legitimate users; without a published false-positive rate or a fast, low-friction appeal mechanism, those users simply lose service and have to fight their way back through the same operator bureaucracy that failed to verify them in the first place.
DoT's own record on proportionality here is mixed. In November 2025, the department ordered handset makers to pre-install a non-deletable Sanchar Saathi app on every new phone sold in India, with a 90-day compliance window (News on Air).
That mandate was withdrawn within days after digital-rights groups objected that forcing a non-removable government app onto every device raised consent and surveillance concerns disproportionate to the fraud problem it targeted. The reversal is a point in DoT's favor — it shows the department can course-correct under scrutiny — but it also means the same institution now running a 14-million-connection disconnection program at algorithmic speed has, within the last eight months, needed public pressure to recognize when it had overreached. That is precisely the track record that argues for building transparency into ASTR now, not after the next controversy.
The Proportionate Path
None of this requires dismantling ASTR or slowing down legitimate fraud enforcement — telecom-enabled fraud is a real and growing harm, and automation is the only realistic way to police a subscriber base of over a billion connections. What it requires is publishing what a periodic transparency report would show: how many flagged connections were false positives, how re-verification is made accessible to subscribers without smartphones or digital documents, and what appeal window exists before a livelihood-critical number goes dark. Algorithmic enforcement that scales this well should be judged by the same standard as any other exercise of state power at scale — not just whether it catches fraud, but whether it can show its work.