Savi Security, founded by two former tech executives, launched an AI-powered app on Tuesday to detect and block sophisticated voice-cloning and text scams targeting consumers. The startup was inspired when Patrick Coughlin's mother nearly fell for a fake kidnapping call using AI to spoof his sister's voice and number. With the FTC reporting $3.5 billion（約5600億円） in imposter-scam losses in 2025 — triple 2020's amount — and voice cloning now possible from just three seconds of social media audio, the app fills a gap in real-time consumer protection.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Brothers Patrick and Ryan Coughlin, both with senior tech careers, founded Savi Security and just raised $7 million（約11億円） in seed funding to protect consumers from AI-generated scams. The company is launching an iPhone and Android app on Tuesday that screens texts, voicemails, and incoming calls for fraud, and can add a live agent to monitor suspicious calls in real time.
Why it matters
AI tools have made sophisticated impersonation scams financially viable for criminals targeting everyday people. The FTC reported that people collectively lost $3.5 billion（約5600億円） to imposter scams in 2025, triple the amount in 2020. Patrick Coughlin's mother nearly fell for a call spoofing his sister's voice and number with a fake kidnapping threat — a case that inspired the startup. Cheap access to voice-cloning and generative AI tools means criminals can now scale attacks that were previously only worth targeting wealthy enterprises or governments.
What to watch
Savi charges $8 per month, discounted to $63 per year, and covers an entire family with no cap on the number of users added to one plan. The company tested its scam-detection model via a free website called Scam Wise, which collected 50,000 submissions in its first four months with roughly 10,000 new submissions per week.
Patrick Coughlin's personal brush with AI-powered fraud — his mother's nearly-successful encounter with a voice-cloned kidnapping hoax — highlights a fundamental shift in the economics of consumer scams. Before generative AI became cheap and widely accessible, the research, voice-spoofing technology, and operational overhead required to impersonate someone made such attacks viable only against high-value targets like enterprises or governments. Now, a criminal can clone a voice from three seconds of publicly available social media audio and deploy it at scale against millions of ordinary people, with negligible cost. The FTC's report of $3.5 billion（約5600億円） in imposter-scam losses in 2025 — triple the 2020 figure — signals that this economic threshold has already been crossed.
Savi's approach mirrors the evolution of consumer antivirus software: deploy AI detection and real-time intervention matching the sophistication of the threat itself. The company's testing phase via Scam Wise generated 50,000 submissions and a growing weekly influx of roughly 10,000 more, suggesting both that the problem is visible and that consumers are willing to share suspicious content for analysis. The live-call monitoring feature, in particular, attempts to catch scams during the moment of highest vulnerability — when a caller claims an emergency and demands immediate payment. Pricing the service as a family plan with no per-user cap may reflect the founders' recognition that fraud victimization often clusters within households and social networks, and that protective value is highest when extended to vulnerable relatives.
No discussion yet for this article
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
