The billions of dollars lost annually to imposter scams are no longer just the work of sophisticated state actors or organized crime rings. Cheap, powerful generative AI has democratized the art of the grift, making it possible for nearly anyone to clone a voice from a three-second social media clip or spoof a family member\u2019s phone number with terrifying accuracy. In response to this new reality, Savi Security, a startup founded by brothers Patrick and Ryan Coughlin, launched its consumer app on Tuesday, promising a real-time defense against AI-powered voice and text scams.<\/p>\n
The company\u2019s origin story is a stark warning of the threat\u2019s potency. Patrick Coughlin, a former national cyber defense professional and a senior executive at Cisco and Splunk<\/a>, received a distraught call from his mother roughly two years ago. She had just received a phone call appearing to come from her daughter\u2019s number. A man stated he had kidnapped the daughter. The mother then heard her daughter\u2019s voice screaming for help, followed by a demand for a $1,200 ransom, referencing the local Walmart her daughter frequented.<\/p>\n Fortunately, the mother retained enough composure to call her daughter directly and discovered she was safe. The entire ordeal was an AI-generated deepfake. The incident horrified Coughlin, who realized the sophisticated attack techniques he had defended Fortune 500 companies and government agencies against were now being weaponized against ordinary consumers. The financial barrier to entry for these scams had collapsed, making them viable for a much broader set of malicious actors.<\/p>\n Savi Security\u2019s solution is to deploy the same transformative technology\u2014artificial intelligence\u2014as a defensive shield. The startup raised $7 million in a seed round led by Acrew Capital to develop a consumer app that screens text messages, emails, and voicemails for malicious intent. However, its most significant feature is an AI agent<\/a> that can listen in on a live phone call. When a conversation feels suspicious, a user can invite the app’s agent to join the line. It analyzes the call in real time for behavioral cues and linguistic tells that indicate a scam is in progress.<\/p>\n Prior to the app\u2019s full release, the brothers validated their detection model with a free, anonymous tool called Scamwise. The site allows users to upload suspicious messages and files for analysis and has already processed over 100,000 submissions. This steady stream of live data has been crucial for training Savi\u2019s detection model, which currently uses Google\u2019s Gemini but is built on a flexible gateway that allows for integration with other specialized voice-detection models as needed.<\/p>\n Savi Security has adopted a unique pricing model aimed at protecting an entire household. For $8 per month or $63 per year, a single account can cover an unlimited number of users\u2014spouses, children, parents, and even that vulnerable uncle who always falls for tech support calls. This design acknowledges that scammers often target the most trusting members of a family, and the primary account holder can manage protection for everyone else.<\/p>\nHow Savi Security Uses AI to Fight AI Scams<\/h2>\n
Pricing, Accessibility, and the Bigger Picture<\/h2>\n