DetectifAI builds on-device AI to spot deepfake voices

After a deepfake voice tricked her grandfather into paying a ransom, Tarini Padmanabhuni founded DetectifAI, a San Francisco startup developing AI models designed to identify fake voices on smartphones.

Padmanabhuni says the company builds compact models to run within a phone’s operating system. Its goal is to assess voices during calls, in voice messages and in other audio without sending the audio off the device. DetectifAI is initially selling its software development kit (SDK) to phone makers, which could integrate it as a built-in feature. The company also plans to license the technology to businesses and fraud-prevention firms.

According to Padmanabhuni, DetectifAI already has early revenue and processes more than 100,000 calls a month for financial institutions in India. AI voice agents use the service for debt collection and loan-document follow-ups, with deepfake detection and speaker verification on each call. She declined to identify the customers, citing confidentiality agreements.

The company also ran a small WhatsApp beta test in which users forwarded suspicious voice notes for assessment. Padmanabhuni says one tester whose relatives had been scammed told her they would pay for the service.

DetectifAI has raised a small seed amount from investors Josh Constine and Manohar Kamath, according to the article. It is among the startups selected to compete in TechCrunch’s Startup Battlefield at TechCrunch Disrupt, scheduled for October 13 to 15 in downtown San Francisco.