Detect if AI Develops Smartphone AI to Combat Deepfake Voice Scams

Tarini Sai Padmanabhuni founded San Francisco-based DetectifAI after a distressing incident two years ago when her grandfather was scammed by a caller who used AI to mimic the voice of his brother, demanding a ransom for a kidnapping that never occurred. This event highlighted the urgent need for reliable tools to detect synthetic voices.

DetectifAI builds compact AI models designed to run directly on smartphones, enabling real-time detection of synthetic or manipulated speech and verifying whether the speaker matches the claimed identity. This on-device approach aims to provide faster, more private protection by analysing calls and voice messages without sending audio data to the cloud.

The company targets phone manufacturers to embed its technology into operating systems, offering an infrastructure-level solution rather than requiring users to install separate apps. DetectifAI’s models can also be deployed via cloud APIs, private customer environments, or telephony edge systems.

Currently, DetectifAI processes more than 100,000 calls per month for unnamed financial institutions in India. These calls, handled by AI voice agents for debt collection and loan follow-ups, incorporate deepfake detection and speaker verification to prevent fraud.

Voice deepfake scams are a growing global problem. The FBI’s 2025 Internet Crime Complaint Center report documented 22,364 complaints involving AI-based fraud, with adjusted losses nearing $893 million. Older adults are frequently targeted with scams impersonating family members during emergencies.

DetectifAI’s technology was independently evaluated in a September 2026 benchmark by Podonos, which tested 23 audio deepfake detection systems on over 4,500 files. DetectifAI achieved 94.47% accuracy and was the fastest commercial system on the leaderboard, reporting a vendor-measured average latency of 24 milliseconds per audio clip.

The startup has raised seed funding from investors including Josh Constine and Manohar Kamath, though details remain undisclosed. DetectifAI was also selected to compete in the 2026 TechCrunch Startup Battlefield event in San Francisco this October, underscoring its growing profile in AI security.

While some claims about customer scale and funding rely on founder statements without independent confirmation, DetectifAI’s approach addresses a critical gap in voice authentication technology by enabling on-device, real-time detection of voice deepfakes to better protect individuals and enterprises worldwide.