Why Telco Giants are Combatting Deepfake Voice Scams with AI

Five major telecommunications operators have launched a joint AI venture to combat the rising threat of deepfake voice fraud, which is costing the industry US$41bn annually and affecting millions of consumers worldwide.

While AI-generated content has gained notoriety for its proliferation on social media platforms such as X, the manipulation of real voices through AI presents an equally serious threat to telcos and their customers.

According to a survey of approximately 12,000 consumers conducted by Hiya across the US, UK, Canada, France, Germany and Spain, one in four American consumers received deepfake voice calls in 2024.

The financial implications could be substantial, with global telecom fraud losses reaching US$41.82bn, according to the Communications Fraud Control Association (CFCA) Fraud Loss Survey Report.

These calls often create panic, pressuring recipients to act quickly before they can assess the legitimacy of the request.

The consequences extend beyond immediate financial loss to include identity theft and long-term fraud affecting individuals and businesses alike.

Syntelligence AI – founded by Deutsche Telekom, e&, Singtel Group, SK Telecom and SoftBank – is deploying AI training datasets created by the five telco giants to address some of the most significant threats facing telco networks today.

Combatting AI-generated threats

Syntelligence AI is targeting threats at network level rather than on individual devices, beginning with its Trust Platform.

The tool addresses the declining user trust in calls from unknown numbers and provides effective screening of spam callers before they reach consumers.

Syntelligence AI has also developed a Security Shield platform, described as a real-time defence model which detects and blocks scam calls before they reach people.

By leveraging network-level data and AI analysis, the system could identify fraudulent calls earlier in the communication chain.

Prateek Choudhary, CEO of Syntelligence AI, says: "The telecommunications industry is entering a new phase in user experience and operational efficiency, enabled by rapid advances in AI.

"It's a significant moment which offers telecom companies a real opportunity to shape what comes next. Building AI-powered services that strengthen their role will deepen customer value and set a new standard for network experience for years to come."

Four-stage AI protection model

At MWC 2026 in Barcelona, Prateek and Deutsche Telekom Vice President Jan Hofmann took part in a panel discussing how the Global Telco AI Alliance is rewriting the rules of network trust.

Prateek said: "We think that AI is the best way to actually solve the problem which is being made worse by AI. What we are doing here is creating a platform called the Trust Platform, which protects you from the first time the phone rings until the end of the call, the entire lifecycle of the call."

The Trust Platform uses four stages of AI-assisted protection throughout the call lifecycle.

Explaining the first stage, Prateek says: "Before the phone rings, the networks have a lot of information, we are leveraging that to determine whether or not the call is a scam or not – so what we call the scam risk assessment."

Telecoms use network signals to conduct this assessment, including roaming status, traffic patterns such as the number of people the phone number has called in the last 24 hours, and crowd intelligence gathered across the network.

The second stage activates when a call rings from an unknown number.

A person can decide to send that call to an AI assistant, which Prateek describes as "basically understanding the intent, understanding why a person is calling. It gathers all the information, summarises it, and lets the person know whether the call was a scam or not". 

If people decide to take the call themselves, the third stage monitors the conversation in real time.

"We monitor the call and see if it is evolving in to the scam call, and we give the right alerts to make people aware that the call could be a scam. Then we take the feedback," Prateek explains.

Proactive intelligence over reactive filtering

In a LinkedIn post, Syntelligence AI states: "By training proprietary AI models on telco-scale data, we are identifying intent before the phone even rings. We are moving from reactive filters to proactive intelligence."

This shift from reactive to proactive defence could represent a significant evolution in how telecommunications networks approach security.

Rather than responding to threats after they emerge, AI models trained on vast datasets from multiple global operators could potentially identify patterns and intentions before fraudulent calls reach consumers.

According to a survey from Hiya, the volume of AI deepfake voice call scams has risen at a compound annual rate of 16% since 2023.

As the AI call assistant solution is rolled out across telco networks, including the five which founded Syntelligence AI, networks could become more secure and less prone to voice call scams.

With scams leading to financial fraud or identity theft affecting all individuals, not just telcos, Syntelligence AI's technology has the potential to become a global solution for telecommunications operators worldwide.