Savi, a consumer protection startup, launched an app this week designed to detect and block AI-generated scams that impersonate relatives in distress. The company raised $7 million in seed funding ahead of the Tuesday release on iPhone and Android.
The app targets a growing threat: voice cloning and deepfake technology that criminals use to demand ransom from family members. In these scams, attackers synthesize a child's or spouse's voice claiming they've been kidnapped and need money immediately. The psychological pressure and speed of these attacks make them effective against vulnerable targets.
Savi's technology analyzes incoming calls and messages to identify hallmarks of synthetic media. The app examines audio characteristics, communication patterns, and behavioral signals that differ from legitimate contacts. When the system detects a likely AI-generated scam, it alerts the user and can block the call or message.
The startup operates in a crowded defensive space. Phone carriers like AT&T and Verizon offer basic spam filtering, while security firms including Truecaller provide broader fraud detection. Savi differentiates by focusing specifically on deepfake and voice cloning threats rather than general spam.
The $7 million seed round reflects investor confidence in consumer protection against emerging AI threats. As voice synthesis tools become cheaper and more accessible, law enforcement agencies warn that synthetic media scams will accelerate. The FBI documented a spike in these incidents in 2023, with victims losing millions.
Savi's app represents one approach to the detection problem. Other researchers develop watermarking systems to mark authentic media and fingerprinting tools to identify synthetic audio. The most effective defense likely combines detection tools with public awareness and rapid reporting mechanisms.
The startup's success depends on its detection accuracy. False positives that block legitimate calls undermine trust. False negatives that miss actual scams defeat the purpose. As scammers adapt their