A grandfather paid a ransom after a voice mimicking his brother said he had been kidnapped, according to an account given to TechCrunch by his granddaughter, DetectifAI founder Tarini Padmanabhuni. TechCrunch described the voice used in the scam as a deepfake, or an AI-generated imitation.
The incident happened about two years before TechCrunch reported the story on 28 September 2026. After making the payment, the grandfather discovered that his brother had been somewhere else and knew nothing about the supposed kidnapping.
The publication did not say how the scammers obtained recordings of the brother's voice or describe any independent forensic examination confirming how the imitation was created.
Padmanabhuni later founded San Francisco-based DetectifAI, which develops technology designed to detect synthetic speech and verify speakers. She told TechCrunch that what stayed with her was how difficult it had been for her grandfather to tell that the voice was not really his brother.
How the Fake Kidnapping Scam Worked
The scam combined two powerful elements of impersonation fraud: a familiar-sounding voice and an apparent emergency.
According to TechCrunch, Padmanabhuni's grandfather believed he was speaking directly to his brother when the voice indicated that he had been kidnapped and that a ransom had to be paid for his release.
The circumstances resemble tactics authorities describe in virtual-kidnapping and family-emergency scams. Fraudsters create a false crisis and pressure targets to act before they can independently check whether their relative is actually in danger.
The FBI has warned that virtual kidnappers exploit fear and urgency to obtain ransom payments before victims can verify what has happened. In a 5 December 2025 alert, the agency warned that criminals were also using manipulated photos and videos as supposed proof of life.
The Federal Trade Commission similarly warns that scammers can use a short audio sample and voice-cloning software to imitate a relative during a fake emergency.
AI-Related Scam Complaints Carry Significant Financial Losses
The FBI's 2025 Internet Crime Report recorded 22,364 complaints reporting AI-related information, with $893,346,472 in adjusted losses associated with those complaints.
Those figures require an important qualification. They do not mean investigators determined that AI directly caused all $893 million in reported losses, nor do they represent losses specifically from voice-cloning scams.
The same report said victims claimed more than $5 million in losses from 'distress' scams in 2025. The FBI said this category includes grandparent scams in which voice-cloning technology can be used to mimic relatives or close friends facing supposed emergencies.
People aged 60 and over filed 201,266 complaints with the FBI's Internet Crime Complaint Center in 2025 and reported $7.748 billion in losses. Those figures cover numerous types of cyber-enabled crime and should not be interpreted as losses specifically caused by AI voice scams.
Grandfather's Experience Led to Voice-Detection Startup
DetectifAI says it is developing compact AI models designed to detect synthetic voices and verify whether speakers are who they claim to be.
Padmanabhuni told TechCrunch that the company's technology is designed to run directly on devices rather than relying solely on remote cloud servers. DetectifAI aims to allow phones and other products to assess audio during calls, voice messages and other interactions without sending the audio off the device.
Padmanabhuni also told TechCrunch that the startup already generates early revenue and handles more than 100,000 calls a month for financial institutions in India.
She said those calls are placed by AI voice agents handling debt collection and loan-document follow-ups, with deepfake detection and speaker verification used on every call. She declined to identify the institutions, citing confidentiality agreements.
Those commercial figures remain claims attributed to the founder rather than independently disclosed customer data.
How Families Can Protect Themselves
Authorities warn against treating a familiar-sounding voice as proof of identity.
The FBI recommends establishing a family code word and attempting to contact the supposed kidnapping victim before considering any ransom demand. It also warns that virtual kidnappers create a false sense of urgency to stop targets from questioning whether the story makes sense.
The FTC similarly advises people to hang up and contact the relative using a phone number they already know is genuine. If the person cannot be reached directly, it recommends checking with another family member or friend rather than immediately acting on the caller's demands.
Voice-cloning technology can make impersonation more convincing, but the underlying tactic remains familiar: fear, urgency and pressure to pay before checking the facts. In cases involving demands for money or sensitive information, a recognisable voice should therefore be treated as one signal, not proof that the caller is who they claim to be.
