AI Automation in Telecom: Ensuring Accountability and Public Trust| India AI Impact Summit 2026

Contents

Executive Summary

This panel discussion from the India AI Impact Summit 2026 explores how AI-driven automation in telecommunications can balance innovation with trust, accountability, and public protection. The panelists—representing regulators, R&D institutions, service providers, and international standards bodies—present concrete use cases, regulatory frameworks, and collaborative approaches to combat fraud while maintaining customer privacy and network integrity. The session emphasizes that responsible AI deployment requires human oversight, cross-sector collaboration, and standardized incident reporting mechanisms.

Key Takeaways

AI-driven fraud prevention at scale (2.1M+ disconnections, 15M spoofed calls blocked daily) is proven, operationalized, and now a benchmark for global telecoms—but requires human oversight, zero-tolerance for false positives, and clear customer communication.
A standardized, voluntary AI incident database (TEC schema) creates a feedback loop for continuous improvement without mandates—analogous to CERT for cybersecurity—enabling regulators, industry, and researchers to prevent recurring AI failures.
Privacy-by-design, certified compliance (ISO 27701, DPDP), and consolidated data architecture are no longer nice-to-have; they are foundational to building and maintaining customer trust in AI-driven services.
Disaster management systems prove AI's high-impact public utility: geotargeted, real-time alerts via cell broadcast can save thousands of lives, and India's model is being adopted internationally to align with UN SDGs.
Innovation in AI (for fraud, security, efficiency) is faster when powered by collaboration across borders, sectors, and regulators—not when constrained by defensive regulation—but requires policy support, sandboxes, and open governance frameworks.

India AI Impact Summit 2026

Key Topics Covered

AI in Fraud & Spam Detection: Real-world applications of AI to identify and block fraudulent connections, SIM spoofing, and scam calls at scale
Customer Trust & Accountability: Role of transparency, human-in-the-loop controls, and privacy-by-design in maintaining public confidence
Standardization & Incident Reporting: World's first AI incident database schema and taxonomy developed by India's Telecom Engineering Centre (TEC)
Network Security & Cyber Defense: Using AI to counter AI-driven cyber attacks on critical telecom infrastructure
Disaster Management: AI-powered early warning systems for cyclones, floods, and emergency alerting (cell broadcast, geotargeting)
Cross-Border Collaboration: International cooperation on data sharing, scam prevention, and regulatory alignment
Enterprise AI Architecture: Consolidation of siloed AI systems toward unified, privacy-compliant platforms
Cost Optimization & Skill Development: Infrastructure costs, upskilling requirements, and AI-aided AI efficiency improvements

Key Points & Insights

Massive Scale of Fraud Prevention: India has disconnected 2.1 million fraudulent numbers using AI-based tracking; C-DOT's FraudPro identified and removed hundreds of thousands of duplicate SIM connections (Jamtara-style sim factories). Government's Sancharati app has 18+ million downloads and 250 million+ website hits, demonstrating public trust in AI-driven consumer protection.
Real-Time Decision-Making at Millisecond Scale: Spoofing call blocking systems must make zero-error decisions within milliseconds as calls hit network gateways. The challenge is preventing actual legitimate calls from being blocked while stopping 15 million spoofed calls per day—now neutralized through rigorous testing and AI-powered filtering.
AI as Shield vs. Regulation as Fence: GSMA argues that regulation moves slower than scammers; innovation through AI and cross-sector collaboration is more effective than restrictive rules. Case studies from 40+ operators across Asia Pacific demonstrate that AI-driven anti-scam strategies succeed without mandates.
Privacy-by-Design & Certified Compliance: Leading TSPs (like Vodafone) are certified to ISO 27701 and DPDP (Data Protection) standards, implementing privacy from the ground up rather than as an afterthought. Consolidating siloed AI data repositories into secure, unified platforms reduces attack surface and improves governance.
Volunteer-Based Incident Reporting Sets Precedent: TEC's new standard (issued November 2025) establishes a 30-field schema for AI incident reporting—the first of its kind globally. Voluntary adoption allows service providers and developers to learn from failures and refine models iteratively, similar to how computer incident reporting evolved post-CERT.
Disaster Management as Public Good: AI-driven early warning systems using geotargeting saved zero lives in Odisha's 2024 cyclone (vs. thousands in 1999 cyclone). Cell broadcast technology (3GPP) paired with sensor fusion (IMD, CWC, weather alerts) reaches all users in at-risk zones—even tourists—within seconds, aligning with UN's Early Warning for All initiative by 2027.
Human-in-the-Loop is Non-Negotiable: While customers may not interact directly with AI models, they are affected by outcomes (outage management, service continuity, grievance handling). Clear, proactive communication and human oversight of automated decisions remain critical to trust; systems must not "run away with their own decisions."
Data Siloing Undermines Efficiency & Security: Enterprise AI adoption initially created 20–30 isolated data repositories and infrastructure silos—each function (HR, network ops, billing) building its own LLM and GPU clusters. Industry is now consolidating toward unified data platforms with purpose-built APIs, reducing cost, complexity, and security risk.
Infrastructure Costs Dominate (80–90% of AI Budget): Skill shortages are secondary; the main challenge is GPU and compute infrastructure. However, upskilling AI specialists (growing from 30 to 3,000 roles post-AI adoption) and human oversight roles remain critical investments that cannot be fully automated.
Cross-Border Collaboration & Open Gateway APIs Drive Innovation: GSMA's four-pillar approach (network security, ecosystem exposure via APIs, customer-facing services, digital skills) requires regulatory sandboxes and policy support. Single entities (including operators) lack complete information needed to prevent scams; collaboration across sectors, jurisdictions, and data-sharing frameworks is essential.

Notable Quotes or Statements

Julian Gorman (GSMA): "AI should be used as a shield and enabler rather than using regulation as a fence and barrier. Scammers are not bound by geography or laws; regulation cannot move as fast."
Julian Gorman (GSMA): "No single entity, especially no mobile operator, has all the information. If operators arbitrarily block SIM cards, you risk the Optus outage scenario where people died because they couldn't call emergency services."
Dr. Raj Kumar (C-DOT): "Disconnected 70 lakh [7 million] connections via Sancharati—people themselves initiated disconnection. This shows the power and popularity of AI-driven customer protection."
Dr. Raj Kumar (C-DOT) (on disaster management): "Death toll in Odisha cyclones: thousands in 1999, zero in 2024. Cell broadcast + AI-driven geotargeting reaches all users at risk, even tourists."
Matan Babu (TSP representative): "The power of AI is not in reducing human involvement; it's in upskilling humans appropriately. We cut 10,000 employees but grew AI staff from 30 to 3,000."
Mr. ST Abbas (TEC): "This standard is not mandatory, just as CERT wasn't mandatory initially—but once incidents started being reported, the mechanism became invaluable."
Moderator (Dr. MP Tangira): "Customers may not interact with AI directly, but they are affected by its outcomes. Clear, proactive communication and human oversight remain critical to trust."

Speakers & Organizations Mentioned

Speaker/Role	Organization
Julian Gorman	GSMA, Head Asia-Pacific
Dr. Raj Kumar	C-DOT (Centre for Development of Telematics), India
Matan Babu Kashilingan	Vodafone India (representing TSPs)
Mr. ST Abbas	TEC (Telecom Engineering Centre), Senior DDG & Head
Dr. MP Tangira	Moderator (affiliation not fully specified)
Chairman Tai	TRI (Telecom Regulatory Authority of India)
Shri Ak Jha	Principal Advisor, TRI

Key Organizations Referenced:

GSMA (Global System for Mobile Communications Association)
C-DOT (Centre for Development of Telematics, India)
TEC (Telecom Engineering Centre, India)
TRI (Telecom Regulatory Authority of India)
RBI (Reserve Bank of India)
Meta, Google, TikTok, AWS (cross-sector anti-scam task force members)
Virginia Tech, GSMA Foundry (research partnerships)
ITU (International Telecommunication Union)
UN (UN Early Warning for All initiative by 2027)
Vodafone India Limited
Bharti Net (network operator)

Technical Concepts & Resources

AI/ML Models & Platforms

FraudPro: C-DOT's image-matching + demographic AI system for detecting duplicate SIM registrations (Aadhaar, driver's license duplication)
Digital Intelligence Platform (DIP): Analyzes 87 crore (870 million) mobile numbers; used by RBI-mandated Financial Risk Indicator (FRI) to assess transaction risk
Chaksh: Crowdsourcing platform for reporting fraudulent/spam calls; integrated with Sancharati app
Sancharati App: 18M+ downloads, 250M+ web hits; uses fuzzy AI and fuzzy logic to identify all connections under a user's name without requiring additional details beyond OTP verification
AI-driven Cyber Security Solution: Fully AI-based for defending against AI-driven cyber attacks (no longer human-only threats)
Cell Broadcast (3GPP): Standardized disaster alerting; broadcasts common messages to all users in a geofenced area (cyclone, flood, tsunami alerts)
Network Management System (NMS): AI predicts network failures, router/node misbehavior; deployed in Bharti Net 1 & 2

Regulatory & Standards Frameworks

TEC AI Incident Database Schema & Taxonomy (November 2025): 30-key-field voluntary standard for reporting AI incidents; categorizes by incident type (network description, security breach, AI mismanagement), affected system (core, RAN, edge, IoT), severity (critical, high, moderate, low), and root cause
ISO 27701: Privacy Information Management Systems standard
DPDP (Data Protection): India's Data Protection law and compliance framework
Open Gateway APIs: GSMA program for standardized data-sharing interfaces
GSMA Cross-Sector Anti-Scam Task Force: 39+ organizations from 17 countries collaborating on scam prevention; working on proof-of-concept for cross-border data sharing (Southeast Asia)
United Against Scams: GSMA global initiative

Disaster Management Systems

IMD (India Meteorological Department), CWC (Central Water Commission), DGSC (Disaster Management Authority): Alert-generating agencies integrated via APIs
SDMA (State Disaster Management Authority): Alert dissemination agencies
Early Warning for All (UN): Target alignment by 2027

Research & Publications

ITU report on cell broadcast disaster management (India's model)
GSMA proof-of-concept on data sharing for scam prevention (Virginia Tech, GSMA Foundry partnership)
40+ case studies of operator-led anti-scam strategies (no regulation) across Asia Pacific

Key Metrics & Data Points

2.1 million fraudulent connections disconnected (nationwide AI-based tracking)
70 lakh (7 million) connections self-disconnected via Sancharati verification
15 million spoofed calls per day (now neutralized)
87 crore (870 million) mobile numbers analyzed in DIP platform
18 million+ Sancharati app downloads
250 million+ Sancharati website hits
Zero deaths in Odisha 2024 cyclone (vs. thousands in 1999, pre-AI system)
80–90% of enterprise AI budget spent on infrastructure (compute, GPU, storage)

Structural Observations

Panel composition balanced across regulators (TRI), R&D (C-DOT, TEC), industry (GSMA, TSPs), ensuring diverse perspectives
Session design prioritized action over theory: real case studies (FraudPro, Sancharati, disaster alerts) over abstract frameworks
Recurring theme: Human oversight and collaboration > automation + control; innovation > restriction
Global-local nexus: India positioned as a telecom superpower with responsibility to export standards and solutions globally while learning from international best practices
Emphasis on voluntary mechanisms (incident reporting, open APIs) over mandatory top-down regulation