Savi Security's $7M Bet on AI Voice-Clone Detection

Savi Security released a consumer app on iOS and Android on July 7, 2026, designed to detect and defend against AI-synthesized voice and video scams — particularly the "kidnapping ransom" variant, where a scammer uses a cloned voice to impersonate a family member in distress and demand urgent payment TechCrunch. The company announced a $7 million seed round led by Acrew Capital, with participation from Magnify Ventures, TTCER, and Resolute Ventures TechCrunch.
Savi was founded by brothers Patrick and Ryan Coughlin. Patrick spent the past several years at Cisco as senior vice president of security products, a role he took on after Cisco's 2024 acquisition of Splunk. Before that, he founded TruSTAR, a cloud security startup that Splunk acquired in May 2021 for roughly $82 million TechCrunch. Ryan's background is in consumer product design, with prior roles at Apple and Spotify.
The founding story has a direct origin: an AI-generated voice scam targeting their own mother, in which someone using a synthetic voice claiming to be a family member demanded ransom. As voice-cloning technology has become cheaper and easier to use — now requiring only a few seconds of audio sample — this class of scam has grown more common and increasingly convincing.
The pairing of founders carries real significance. Patrick's expertise lies in enterprise-grade security architecture — the kind of threat detection and analysis built into corporate security operations centers that process hundreds of alerts daily. Ryan brings consumer product design, the discipline of making something intuitive for someone receiving a panicked phone call, not a trained analyst. Savi's challenge is translating the behavioral and signal-analysis techniques from enterprise threat intelligence into a product that works for ordinary people in a moment of fear and urgency.
Acrew Capital's lead on the round aligns with the firm's track record in consumer security and fintech-adjacent companies. The syndicate's modest size relative to some recent AI-security funding rounds suggests Savi is still establishing product-market fit at the consumer level rather than scaling enterprise sales.
The problem space is real. Voice and video synthesis tools have become commodified, and phone-based social engineering — in particular, calls claiming a family member is in danger — has been a persistent vulnerability. Law enforcement agencies and consumer advocates documented this pattern as early as 2023, but the quality of synthesized audio has improved substantially in the intervening years, narrowing the gap between a scam call and a legitimate one.
The consumer scam-detection app market is already dense, spanning carrier-level spam filtering, native call-screening in iOS and Android, and various third-party apps with mixed effectiveness. Savi's differentiation, based on available information, rests on specialized detection for AI-synthesized voice and video rather than generic spam or robocall filtering. Whether that focus will hold as threats themselves evolve is something the company will need to prove over time.
The founding narrative functions as more than marketing. It illustrates where the threat has shifted. Enterprise security has spent years building institutional defenses — SOC tools, threat intelligence sharing, zero-trust architectures — against attacks on corporate infrastructure. AI voice scams targeting families operate on a different surface entirely, with no institutional barrier between attacker and victim's phone. A company founded by an enterprise security veteran, prompted by a scam targeting his own family, points directly to that gap.
Savi's immediate test is demonstrating detection accuracy at scale without triggering false alarms that undermine confidence at the exact moment — a real family crisis — when a missed detection would be costliest. Managing that tension between sensitivity and specificity in a high-stakes, real-time consumer scenario is likely the more revealing measure of the company's prospects than the funding figures themselves.


