San Francisco-based startup DetectifAI has developed compact AI models designed to run locally inside smartphone operating systems to detect voice deepfakes instantly. Founded by Tarini Padmanabhuni after her grandfather fell victim to a kidnapping ransom scam using an AI-generated voice clone, the company aims to protect users from audio fraud without requiring their audio to leave the device.
While competitors like Reality Defender, Pindrop, and Microsoft offer deepfake detection that runs on remote cloud servers, DetectifAI builds its compact models specifically for on-device use. The startup is targeting phone manufacturers by licensing its software development kit so that detection can be embedded natively into mobile operating systems. Additionally, the company already generates early revenue by handling over 100,000 calls monthly for financial institutions in India, providing deepfake detection and speaker verification for AI voice agents used in debt collections.
Backed by seed investors Josh Constine and Manohar Kamath, DetectifAI has been selected to compete in TechCrunch's Startup Battlefield event at TechCrunch Disrupt in San Francisco. This technology matters because it offers a proactive, privacy-preserving defense against the millions of dollars lost annually to sophisticated AI-driven voice scams.