Tarini Padmanabhuni founded San Francisco-based DetectifAI after an AI-generated voice deepfake tricked her grandfather into paying a ransom for a kidnapping that never happened. The startup is now positioning itself to bring deepfake detection directly onto smartphones, aiming to protect users from the growing threat of synthetic voice fraud.
What Happened
The incident that sparked DetectifAI’s creation occurred approximately two years ago, when Padmanabhuni’s grandfather received a call from a voice he believed was his brother’s. The caller claimed to be kidnapped and demanded a ransom, which the grandfather paid before discovering his brother was elsewhere and unaware of the scam. “What stayed with me wasn’t the money,” Padmanabhuni said. “It was that he had no way of telling.”
DetectifAI aims to address this vulnerability by developing compact AI models designed from the start to run inside a smartphone’s operating system. Unlike competitors that shrink large cloud models for local use, the company claims its approach allows for instant verdicts on whether a voice is AI-generated during calls and voice messages without the audio ever leaving the device. The core product is a software development kit (SDK) licensed to phone manufacturers, enabling detection to ship as a built-in feature.
Why It Matters
The urgency of this technology is underscored by rising fraud statistics. According to the FBI, Americans lost nearly $900 million to AI-driven scams last year, a 24% increase from 2024. The financial impact is disproportionately felt by older adults, with people aged 60 and older losing twice as much as those aged 50 to 59.
While established players like Reality Defender, Pindrop, Resemble AI, Microsoft Azure AI Content Safety, and Nuance offer deepfake detection, Padmanabhuni argues that their reliance on remote cloud servers prevents phone makers from integrating these tools directly into devices. By focusing on on-device processing, DetectifAI seeks to provide real-time defense against fraud at the point of contact. The company has already generated early revenue, handling more than 100,000 calls a month for financial institutions in India, where its technology supports AI voice agents for debt collections and loan follow-ups with speaker verification.
Padmanabhuni, who began working in machine learning at age 12 and studied cyber-physical systems at Manipal Institute of Technology, has raised a small seed round from investors including Josh Constine and Manohar Kamath. DetectifAI is also one of the startups selected to compete in TechCrunch’s Startup Battlefield at the upcoming Disrupt conference in San Francisco.
The Bottom Line
DetectifAI is leveraging a personal fraud story to launch a niche but critical solution in the fight against AI voice scams. By licensing its on-device detection SDK to phone manufacturers, the startup aims to make deepfake verification a standard, invisible layer of security for mobile users, capitalizing on a market where cloud-based solutions currently leave gaps in real-time protection.