AI-generated voice cloning is adding a new dimension to cyber fraud, enabling criminals to imitate the pitch, accent, rhythm and speech patterns of real people
In January this year, Smita, a school teacher working in Indore, Madhya Pradesh, received a desperate call from someone who sounded exactly like her cousin. The caller claimed a friend had suffered a heart attack and urgently needed money for treatment. Trusting the familiar voice, she transferred INR 97,500 in four transactions. The next morning, she discovered her cousin had never called. Police suspected the fraudsters had used AI-based voice-cloning technology.
In another case, earlier this month in Kolkata, West Bengal, a 67-year-old man allegedly lost INR 95,990 after receiving a call from someone whose voice sounded like his acquaintance, Ajoy. The caller claimed a relative needed urgent money for hospital treatment. The victim transferred the money before discovering that his acquaintance had never made the call.
These cases show how a familiar voice and a fabricated emergency can trigger panic and trust. As Artificial Intelligence becomes more accessible, scammers around the country are making it into a tool for carrying out frauds as victims struggle to distinguish between a true call from someone close who is in trouble and a fraudster.
AI-generated voice cloning is adding a new dimension to cyber fraud, enabling criminals to imitate the pitch, accent, rhythm and speech patterns of real people. Combined with traditional tactics such as urgency, fear, secrecy and emotional manipulation, the technology is making fraudulent calls increasingly convincing.
Bhavishya Pandit, Senior Data Scientist at 66degrees, a an AI-native technology and consulting company, based in Bengaluru, Karnataka, and a Google Developer Expert, says the power of such scams lies in the fact that people instinctively trust familiar voices.
Also Read: AI Impact Summit: Disruptive Technology or Traumatic Asphyxia
“We recognise people by voice before we even process what they are saying, it’s one of the oldest trust signals we have. AI tools now need just a few seconds of audio, often lifted from a reel or YouTube video, to clone that voice convincingly. So when something familiar sounds urgent, the brain reacts emotionally first,” Pandit tells Media India Group.
According to data from the Ministry of Home Affairs (MHA), reported losses from cyber fraud rose from INR 22.9 billion in 2022 to INR 74.65 billion in 2023 and INR 228.46 billion in 2024. The nearly 10 fold increase in reported losses over two years highlights the rapidly escalating financial cost of cybercrime.
By June 30, the government’s Citizen Financial Cyber Fraud Reporting and Management System (CFCFRMS) had helped save more than INR 111.58 billion across over 3.28 million complaints since its launch.
Surveys suggest that AI voice fraud is becoming a significant concern. A 2023 McAfee survey, later cited in an Observer Research Foundation (ORF) analysis, found that 83 pc of Indian victims of AI voice scams suffered financial losses, while nearly half lost more than INR 50,000.
The survey also found that 47 pc of Indian adults had either experienced an AI voice or deepfake scam or knew someone who had, while 69 pc said they could not confidently distinguish an AI-generated voice from a real one.
AI voice scams typically begin with fraudsters collecting a person’s voice from social media videos, podcasts, voice notes or other recordings and using AI to create a similar-sounding voice. They then impersonate relatives in distress or trusted officials to create urgency and panic. Their success lies as much in psychological pressure as in technology, with a study of 4,100 US participants finding that, in the most persuasive AI voice-phishing scenario, up to 36 pc were willing or potentially willing to comply.
Pandit says it is becoming increasingly difficult to distinguish a cloned voice from a real one, though some warning signs may still help. The ‘robot’ vibe unusual pacing, a voice that sounds too flat or slightly unnatural, can be one clue. Strange or unnaturally consistent background noise may be another.
He also advises asking the caller something that only the real person would know. A vague response or repeated answer should raise suspicion. In video calls, even a slight mismatch between lip movements and audio or unnatural blinking can be a warning sign.
Also Read: QR codes becoming new trap for phishing: Planet VPN
Most importantly, Pandit highlights what he calls the Panic Button.
“Scammers love to make you feel like you have to act right now and keep it a secret,” he says. However, he cautions that victims should not rely only on technical glitches or audio clues, as the technology is advancing rapidly.
The government has been expanding its response to the broader cybercrime threat. According to the latest MHA data, the national Suspect Registry had received more than 3.05 million suspect identifiers by June 30, while information on over 3.21 million Layer-1 mule accounts had been shared. The system helped decline or prevent transactions worth approximately INR 257 billion.
More than 1.6 million SIM cards and over 577,000 IMEIs had also been blocked in connection with cybercrime-related activity. India had 459 dedicated cybercrime police stations, according to a 2026 parliamentary reply.
For victims, immediate reporting is crucial. Financial cyber fraud can be reported through the national cybercrime helpline, 1930, or the National Cyber Crime Reporting Portal. Prompt reporting can increase the possibility of stopping or freezing a transaction before the money is transferred through multiple accounts.
The Reserve Bank of India has repeatedly warned users never to share OTPs, PINs, passwords or CVVs over phone calls. They advise people to independently verify emergencies by disconnecting the call and contacting the person through a known number or another communication channel.
Pandit also recommends having a family safe word something personal that a scammer cannot easily find on social media.
“Never, ever send money or give out an OTP or bank info just because a voice sounds convincing. At the end of the day, if someone is rushing you for money or data, it is almost certainly a scam, no matter how much they sound like your mom or your manager,” Pandit adds.
