AI voice detection startup DetectifAI launched

The rapid development of AI technologies brings not only new opportunities but also dangerous tools for cybercrime. To combat one such threat, audio deepfake fraud, San Francisco-based company DetectifAI has unveiled its new developments. The primary motivation for founding this startup was an unpleasant incident in the founder's family, and the technology aims to protect people from voice spoofing. This is reported by Techcrunch.com reports.

According to ixbt.com, two years ago, Tarini Padmanabhuni's grandfather became a victim of scammers. A caller introduced himself as his grandson, claiming he had been kidnapped and that a ransom was needed. The grandfather paid the money without realizing the voice had been faked using AI. This incident focused the founder's attention on personal security and the problem of voice authentication.

According to the FBI, Americans lost nearly $900 million to AI-assisted fraud last year, a 24% increase from the previous year. Seniors over 60 were hit twice as hard as those aged 50–59. This indicates a rapidly growing demand for protective tools in this sector.

AI solution specifically for smartphones

While companies like Reality Defender, Pindrop, Resemble AI, Microsoft Azure AI Content Safety, and Nuance currently exist to detect voice deepfakes, most operate on cloud servers. Padmanabhuni notes that existing products rely on remote servers, and phone manufacturers cannot integrate them directly into devices, leaving the user vulnerable.

Instead of shrinking large cloud models, DetectifAI developed compact AI models from the ground up that can run directly on smartphone operating systems. This solution allows for the immediate detection of whether a voice was AI-generated during calls, voice messages, and other audio recordings, without sending any data outside the device.

Market competition and future plans

The company plans to license its software to phone manufacturers to make it a core part of the device. Currently, the startup serves over 100,000 calls per month for financial institutions in India. These calls are handled by AI agents involved in debt collection and credit verification, utilizing deepfake detection and identity verification functions in every process.

Tarini Padmanabhuni has been active in programming and machine learning since the age of 12. She studied cyber-physical systems at the Manipal Institute of Technology in India and was the youngest team leader for the first driverless race car division in India at the Formula Student international engineering competition.