TRAI mandated telecom operators to share AI-detected spam signals across networks within two hours by February 2026. Operators are resisting this, arguing their AI systems have different spam definitions, leading to "a guess about a guess" when flags are shared. This creates a regulatory deadlock over how probabilistic AI signals should be used for cross-network action.
How We Got Here
The directive came as part of TRAI's ongoing efforts to curb spam and fraud calls plaguing Indian consumers. The February 2026 mandate aimed to leverage AI for faster, coordinated action across all telecom networks.
The Numbers
- The roundtable discussion on AI's impact on spam prevention was held on August 12 in New Delhi under Chatham House Rule.
- Operators use different AI models and parameters, making their spam detection systems disagree on what constitutes a spammer.
- A number flagged as probable spam by one network's AI then travels across others, where it's treated as a certainty, becoming "a guess about a guess."
- Operators want shared signals to inform their decisions, not dictate an automatic block based on another network's unverified flag.
- Government agencies are pushing operators to hand over these probabilistic AI spam indicators to law enforcement for FIRs, which operators resist.
What Happens Next
🇮🇳 Why This Matters for India
For the 80 crore Indian mobile users bombarded daily with spam calls, this technical disagreement between TRAI and telcos means a delayed resolution to unwanted intrusions.
The Take
TRAI overreached by mandating outright blocking based on probabilistic AI. The real solution lies in a federated learning approach where operators pool data, not just raw flags, enabling each telco to improve its own models.
Source:
MediaNama ↗