AI Security
Founder launches on-device deepfake voice detector after scam targets her grandfather
DetectifAI is building AI voice verification into smartphone operating systems to catch spoofed calls before they can fool users—starting with phone manufacturers.
TechCrunch — AI
Tarini Padmanabhuni's grandfather received what seemed like an urgent call from his brother claiming to have been kidnapped and demanding ransom payment. He wired the money without hesitation—only to later discover the voice was artificial, a deepfake engineered to exploit his trust. The real brother was safe the entire time.
"What stayed with me wasn't the money," Padmanabhuni recalls of the incident roughly two years ago. "It was that he had no way of telling."
That personal brush with AI-powered fraud sparked the creation of DetectifAI, a San Francisco startup designed to embed voice authenticity checks directly into phones so users can spot synthetic audio in real time.
The Scale of AI Voice Fraud#
Voice-based AI scams have become a major financial threat. The FBI documented that Americans lost nearly $900 million to AI-driven fraud last year—a 24% surge from 2024. Seniors aged 60 and above were hit hardest, losing roughly twice as much as those aged 50 to 59, making them particularly vulnerable to elaborate impersonation schemes.
A Different Approach to Detection#
The field already contains established players—Reality Defender, Pindrop, Resemble AI, Microsoft Azure AI Content Safety, and Nuance among them—but Padmanabhuni argues they share a critical weakness: their detection systems run on remote cloud servers, meaning phone manufacturers cannot bake them directly into devices. That leaves users with limited on-the-spot protection.
Rather than compress existing large models to fit phones, DetectifAI designed lightweight AI models from the ground up that operate inside a smartphone's operating system. The technology analyzes incoming calls, voice messages, and audio to instantly classify whether a voice is genuine or synthetic—all while the audio remains on the device and never touches external servers.
Building In as a Standard Feature#
The startup is targeting phone manufacturers first, licensing its software development kit (SDK) so deepfake detection arrives as a native operating system capability. Padmanabhuni draws a parallel to AT&T's exclusive carrier partnership with the original iPhone: the first phone maker to ship DetectifAI gains competitive differentiation, and detection eventually becomes an expected feature like screen resolution.
Secondary revenue will flow from licensing the technology to businesses and anti-fraud services. The company already has paying clients—financial institutions in India handling over 100,000 calls monthly, where AI voice agents conduct debt collection and loan follow-ups. Each call is screened for synthetic audio and speaker verification.

