Bharti Airtel launches AI-based solution to detect spam calls and messages

Developed in-house, the solution uses 250 parameters to identify and flag suspicious communication, starting Thursday.

According to Gopal Vittal, Bharti Airtel's managing director and CEO, the proprietary algorithm has already demonstrated a 97% success rate in identifying spam.

The AI solution analyzes real-time factors like caller usage patterns, call/SMS frequency, and duration, cross-referencing this data with known spam patterns.

Among the 250 parameters, factors such as call velocity, frequency of device changes, short-duration calls, and geographical spread help determine whether a call or message is spam.

The system also flags suspected robocalling devices and assesses multiple names linked to the same KYC.

What sets Airtel’s solution apart from platforms like Truecaller is the sheer scale of its network.

Vittal explained that the system features dual layers of protection: one at the network level and another at the IT systems level.

Every day, Airtel’s "AI shield" processes 1.5 billion messages and 2.5 billion calls within two milliseconds, equivalent to managing 1 trillion records in real-time. The solution identifies 100 million potential spam calls and 3 million spam SMS messages daily.

The solution also offers real-time alerts for malicious links sent via SMS by scanning messages against a centralized database of blacklisted URLs.

Vittal highlighted that the system does not read the content of SMS messages but only checks for suspicious links.

The AI solution is also capable of detecting fraud indicators such as frequent changes to device IMEI numbers.

However, some features are still in development, including coverage for feature phones, video calls, and WhatsApp calls, which remain encrypted and outside the solution's reach.

Despite being a work in progress, the AI system provides Airtel customers with robust protection from the growing threats of spam and fraud.