An AI-generated voice cloned his brother, convinced him of a kidnapping, and extracted a ransom. The victim had no way to know the call was fake. That moment, two years ago, set Tarini Padmanabhuni on a path to build detection that works directly inside a phone, before the scammer ever reaches the speaker.
Padmanabhuni is the founder of DetectifAI, a San Francisco startup that builds compact AI models designed to spot synthetic voices during live calls, in voice messages, and across other audio. The company sells first to phone manufacturers, licensing a software development kit (SDK) so that detection ships as a built-in feature of the device's operating system. The audio never leaves the phone.
The market need is measurable. The FBI reports that Americans lost close to $900 million to AI-driven scams last year, a 24% jump from 2024. Adults 60 and older lost twice as much as those aged 50 to 59.
A detection engine that runs on-device
Several companies already sell deepfake voice detection, including Reality Defender, Pindrop, Resemble AI, Microsoft Azure AI Content Safety, and Nuance. But Padmanabhuni points to a gap: those products run in the cloud on remote servers. Phone makers cannot embed them directly into a handset, leaving the person receiving a call with no real-time defense.
DetectifAI takes a different approach. Instead of shrinking a large cloud model to fit on a phone, the team designs small AI models from the start that are compact enough to run inside a smartphone's operating system. The goal is an instant verdict on whether a voice is synthetic, without routing audio to an external server.
A two-track revenue model
The startup's primary product is its SDK, a package of code that other companies license and build into their own products. Padmanabhuni compares the strategy to AT&T's exclusive deal for the original iPhone: the first phone maker to ship DetectifAI gains a feature rivals lack. A secondary revenue stream will come from licensing the technology to businesses and fraud-prevention firms.
DetectifAI already handles more than 100,000 calls a month for financial institutions in India, according to Padmanabhuni. Those calls are placed by AI voice agents handling debt collections and loan-document follow-ups. Every call includes deepfake detection and speaker verification. Padmanabhuni declined to name the customers, citing confidentiality agreements.
The founder's technical roots
Padmanabhuni said she began working in machine learning at age 12. She later studied cyber-physical systems at Manipal Institute of Technology in India, where she became the youngest team lead of what she describes as India's first driverless race car division in the Formula Student engineering competition.
DetectifAI has raised a small seed round from investors Josh Constine and Manohar Kamath, a principal at KM Growth. The company is also a finalist in the Startup Battlefield competition at TechCrunch Disrupt, scheduled for October 13 to 15 in San Francisco.
Why this matters for customer support, finance, and insurance teams
Voice scams and synthetic-audio fraud hit industries that rely on phone-based trust. A finance or insurance agent who cannot verify whether a caller's voice is real faces direct financial liability. Customer support teams become unwitting entry points for social engineering when they trust a voice that a machine generated. On-device detection shifts verification to the hardware layer, where it can operate before a human agent or an elderly relative ever picks up the call. For professionals managing fraud risk, procurement of voice-security tools, or compliance in voice-channel operations, the move toward embedded detection signals a coming standard in device-level defense.
Your membership also unlocks:
