San Francisco-based startup DetectifAI is developing compact AI models that can detect deepfake voices directly on smartphones, without sending audio to remote servers. Founder Tarini Padmanabhuni started the company after her grandfather was tricked into paying a ransom during a call that used an AI-generated imitation of his brother’s voice.
The call convinced him that his brother had been kidnapped, but the relative was elsewhere and unaware of the scam. Padmanabhuni said the incident showed how difficult it had become for people to determine whether a familiar voice was real.
The problem is growing alongside the availability of voice cloning tools. According to the FBI’s 2025 Internet Crime Report, Americans reported nearly $893 million in losses tied to AI-related scams last year, with older adults among the groups experiencing substantial fraud losses.
Deepfake Detection Runs Directly on the Device
DetectifAI says its models are designed from the beginning to be small enough to run within a smartphone operating system. The goal is to identify AI-generated voices during live calls, voice messages, and other audio without requiring the recording to leave the device.
That differs from many competing products that rely on cloud infrastructure. DetectifAI is initially targeting phone manufacturers and plans to license its software development kit so deepfake detection can be built directly into devices.
The company’s technology also includes speaker verification, which checks whether a caller matches a claimed identity. DetectifAI says the same underlying models can be used for deepfake detection, authentication, transaction confirmation, account recovery, contact-center fraud, and voice agent security.
DetectifAI has also published research on deploying audio deepfake detection across GPU, CPU, and Android devices, reporting that 99.65% of handset clips in its tests were processed faster than real time.
Financial Institutions Are Already Using the Technology
Padmanabhuni said DetectifAI already generates early revenue and processes more than 100,000 calls each month for financial institutions in India. Those calls involve AI voice agents handling debt collection and loan-document follow-ups, with deepfake detection and speaker verification applied to each interaction.
The startup has raised a small seed round from investors Josh Constine and Manohar Kamath of KM Growth. Padmanabhuni declined to identify customers because of confidentiality agreements.
DetectifAI is also one of the startups selected to compete in TechCrunch’s Startup Battlefield at Disrupt 2026. The company plans to generate additional revenue by licensing its technology to businesses and fraud-prevention companies alongside its work with smartphone manufacturers.
Featured image credits: Magnific.com
For more stories like it, click the +Follow button at the top of this page to follow us.