The Telecom Regulatory Authority of India (TRAI) directed telecom operators in February 2026 to share AI-detected spam signals across networks within two hours, aiming to curb spam calls more effectively. However, operators use different AI models to identify spam, leading to inconsistencies in flagged numbers, according to a MediaNama roundtable held on August 12 in New Delhi.
At the roundtable, participants discussed how each operator builds its own spam-detection system with unique parameters, resulting in divergent outputs. One speaker noted that a number flagged as spam by one operator might not be flagged by another, making the shared lists of spam numbers non-identical. This discrepancy complicates the enforcement of spam blocking across networks, as the flagged numbers from one operator may not match those from another.
The directive to share AI spam signals within a two-hour window aims to create a coordinated response to spam calls. However, the lack of standardization in AI models means that flagged numbers are treated as definitive once shared, even though the underlying detection criteria differ. This situation highlights the challenges regulators face in harmonizing AI-based spam prevention efforts across multiple telecom providers with varying technologies.
TRAI’s February 2026 order requires operators to exchange flagged spam numbers promptly, but the differing AI detection systems mean that a flagged number on one network may not be flagged on another. The roundtable underscored the complexity of implementing a unified AI-driven spam prevention framework across India’s telecom sector.