CIO Insider

CIOInsider India Magazine

Separator

What is Airtel's New AI Spam Call Detection System?

Separator

Beginning the morning of September 26, 2024, Airtel subscribers got a suspected spam banner on their dialer screen when they received a call from a number detected as spam or fraudulent by the telecom giant's new AI spam detection system. SMS with potentially malicious links will likewise be flagged as suspected spam. Spam filter optimization is automatically enabled at no additional cost to both prepaid and postpaid Airtel subscribers. Outside of smartphones, the company is also looking into ways to enable AI-based spam detection for feature phones with automated moderation.

Bharti Airtel’s AI-Powered Spam Detection Solution
Bharti Airtel’s first-of-its-kind AI-powered spam detection solution alerts customers of suspected spam calls and SMS in real-time. The solution is free of charge and works automatically for all Airtel customers without the need to raise a service request or download an app.

Gopal Vittal, Managing Director and Chief Executive Officer, Bharti Airtel, says, “Spam has become a menace for customers. We have spent the last twelve months trying to solve this comprehensively. Today marks a milestone as we launch the country’s first AI-powered spam-free network to shield our customers from the continuous onslaught of intrusive and unwanted communications”.

“Designed as a dual-layer protection, the solution has two filters – one at the network layer and the second at the IT systems layer. Every call and SMS passes through this dual-layered AI shield. In 2 milliseconds, our solution processes 1.5 billion messages and 2.5 billion calls daily. This is equivalent to processing 1 trillion records on a real-time basis using the power of AI. Our solution has successfully identified 100 million potential spam calls and 3 million spam SMS originating daily. For us, keeping our customers secure is a burning priority.” adds Gopal.

How Does It Work?
Bharti Airtel's spam detection solution is designed as a dual defense. There are two filters, one at the network layer and one at the IT system layer. All calls and SMS pass through this dual-layered AI shield. The solution processes 1.5 billion messages and 2.5 billion calls every day in 2 milliseconds. This is equivalent to processing 1 trillion records in real-time using the power of AI. The company’s solution successfully identifies 100 million potential spam calls and 3 million spam SMS daily. For us, keeping our customers safe is a top priority.

We are open to doing it. We are open to sharing data where there is legitimate scam or spam. This solution tries to stop scams by making sure that the calls don’t happen because every scam begins with that call

Developed in-house by Airtel's data scientists, this AI-powered solution uses proprietary algorithms to identify and classify calls and SMS as suspected SPAM. It asses the duration of calls and other parameters in real-time. By matching this information against known spam patterns, the system accurately flags calls and SMS as suspected spam.

In addition, the solution also warns customers about malicious links received via SMS. To this end, Airtel has built a centralized database of blacklisted URLs, and all SMS are scanned in real-time by an AI algorithm to alert users not to accidentally click on suspicious links.

The solution can also detect anomalies such as frequent IMEI changes, a typical indicator of fraudulent activity. By layering these protective measures, the company ensures that its customers receive maximum protection against evolving spam and fraud threats.

How is it Different from Apps?
Airtel's AI spam filter was a better solution than Truecaller. That's because such crowdsourcing apps rely on user feedback to detect spam and require an Internet connection to function. 'We don't even know we're using Truecaller because it doesn't recognize the call as a suspected scam. In our case, it detects them in real time. The Calling Name Presentation (CNAP) proposed by the Telecom Regulatory Authority of India (TRAI) would also not work, as many subscribers may not have registered their SIMs in their names.

However, the AI spam filter has certain limitations. This is because 3 percent of calls come from companies like Swiggy, which could be labeled as spam even if they are genuine. A few days ago, Airtel's CEO sent a letter to the heads of rival telecom companies like Jio, Vi, and BSNL regarding the need to share data on B2B subscribers to differentiate between spam calls and legitimate calls from companies.

The proposed measure will not deter the growing number of international spam/scam calls, as the AI spam filter will only apply to Indian networks. According to recent data, more than 85,000 scams targeting Indians have originated from Southeast Asian countries so far this year. Furthermore, OTT communications would not be covered.

Can AI Help Eliminate Spam/Scam Calls and SMS?
In June of this year, TRAI directed all telecom service providers to activate their AI/ML spam detection systems to curb unsolicited communications from unregistered telemarketers (UTMs). The directive specifically instructed telecommunication service providers to devise systems that can constantly evolve to deal with new signatures, new patterns, and new techniques used by UTMs, after a similar directive failed to produce the expected results.

Experts have also long pointed out that AI-based spam detection mechanisms may be inadequate without serious consequences for repeat offenders. Recently, TRAI submitted a discussion paper proposing differential rates for subscribers who make more than 50 calls or send more than 50 SMS per day.

Gopal says, “We are open to doing it. We are open to sharing data where there is legitimate scam or spam. This solution tries to stop scams by making sure that the calls don’t happen because every scam begins with that call.”



Current Issue
Trust Is At The Center of BFSI Transformation