Savi’s app aims to protect consumers from realistic AI scams like kidnappers demanding ransom
Key Takeaways
- Savi Security, founded by brothers Patrick and Ryan Coughlin, has raised $7 million in seed funding led by Acrew Capital, with Magnify Ventures, TTCER, and Resolute Ventures also participating, according to TechCrunch.
- The company is launching its consumer app for iPhone and Android on Tuesday, offering real-time screening of texts, voicemails, and calls, plus a live-call monitoring feature where an agent can listen in during a suspicious conversation, TechCrunch reported.
- The idea grew out of a personal incident: Patrick Coughlin’s mother received a call with a spoofed caller ID and an AI-cloned voice claiming his sister had been kidnapped, an event he described to TechCrunch as the catalyst for the company.
A Personal Scare Becomes a Security Startup
Savi Security was built by Patrick and Ryan Coughlin, two brothers whose prior careers spanned national cyber defense work, Splunk, Cisco, Apple, and Spotify, according to TechCrunch. Patrick Coughlin had been senior vice president of security products at Cisco, a role he reached after Splunk acquired his earlier startup, TruSTAR, for a reported $82 million in May 2021; Cisco later acquired Splunk in 2024, TechCrunch noted.
The company’s founding story centers on an incident roughly two years ago involving the brothers’ mother. She received a call that appeared, via caller ID, to come from her daughter, and during the call she believed she heard her daughter’s voice say she had been taken, followed by a scream and a man demanding $1,200 to prevent harm at a Walmart parking lot the daughter was known to frequent, Patrick Coughlin recounted to TechCrunch. The scammer had spoofed the phone number, mimicked the voice, and referenced a real location tied to the family, making the ruse highly convincing. The mother stayed calm, called her daughter directly, and confirmed she was safe, TechCrunch reported.
Coughlin told TechCrunch that the episode led him to question how the kind of sophisticated attacks he had previously seen aimed at government agencies and Fortune 500 companies were now being turned against ordinary consumers. His conclusion, as relayed by TechCrunch, was that inexpensive and powerful large language models and other generative AI tools have made this level of deception financially viable at the individual level for the first time.
Why Cheaper AI Tools Change the Threat Calculus
Before generative AI became widely accessible, mounting a convincing, personalized scam against a single consumer required significant research and specialized tools, such as voice-spoofing technology, according to the account Coughlin gave TechCrunch. That cost structure meant such attacks were largely reserved for high-value targets like corporations or government bodies, and the defensive technology followed the same pattern. Coughlin told TechCrunch that voice cloning now requires as little as three seconds of audio, which can often be pulled from a public social media post, meaning ordinary moments like recording a child’s sports game and posting it online can supply enough raw material for impersonation.
TechCrunch cited FTC figures showing that people reporting online crimes lost a combined $3.5 billion to impostor scams in 2025, three times the amount recorded in 2020. TechCrunch also referenced 2025 research from Malwarebytes indicating that while older Americans make up the majority of scam reports, Gen Z is frequently targeted through text-based scams and falls for them about 25% of the time. These figures illustrate a broader pattern relevant to anyone active in fast-moving digital economies, including crypto markets, where impersonation, urgent payment demands, and convincing but fabricated communications are already common attack vectors. As AI lowers the cost of producing realistic voice and text deception, the same techniques described by Coughlin to TechCrunch could plausibly extend to fraudulent requests involving wallet transfers or urgent “verify your account” messages, though the source material discussed here focuses specifically on family-impersonation and ransom-style scams rather than crypto-specific incidents.
To build and test its detection systems, Savi first launched a free, registration-free tool called Scam Wise, which allows anyone to upload suspicious texts, photos, or emails for an assessment of whether they are likely fraudulent, TechCrunch reported. According to Coughlin’s comments to TechCrunch, the tool launched about four months before this reporting, received roughly 50,000 submissions, and was growing by about 10,000 or more submissions per week. That data, TechCrunch noted, has been used to train Savi’s scam-detection model, which currently relies primarily on Google’s Gemini but is built on an AI gateway that lets the company incorporate other specialized models, including tools focused on voice detection.
What the App Actually Offers
The paid app launching Tuesday screens incoming texts, voicemails, and calls, a category of feature TechCrunch noted already exists in products such as those from Malwarebytes. Savi’s distinguishing feature, according to TechCrunch, is live-call monitoring: during an active phone conversation a user suspects may be fraudulent, they can add Savi’s live agent as a listener, and the system evaluates behavioral cues in real time to help determine whether the call is a scam in progress.
Pricing is structured around households rather than individual users. TechCrunch reported the service costs $8 per month, or a discounted $63 per year, and that a single subscription covers an entire family with no limit on the number of users who can be added, whether that includes children, a spouse, parents, or other relatives needing extra support. Coughlin framed the broader shift to TechCrunch as one where generative AI has lowered the barrier to committing fraud so significantly that it is drawing in not just organized criminal networks but also individuals who might not otherwise have engaged in deception, describing Savi’s approach as applying AI defensively in real time, in much the way malicious actors already do offensively.
Hype Check
Claim: Savi Security’s app can protect ordinary consumers from realistic AI-generated scams, including voice-cloned kidnapping ransom calls, using real-time detection and live-call monitoring. Reality: TechCrunch’s reporting confirms the $7 million seed round, the Tuesday app launch, the founders’ personal motivating incident, and specific details on Scam Wise’s submission volume and the app’s family-based pricing, but the source does not include independent, third-party verification of the app’s detection accuracy or effectiveness once live in the market. Verdict: Mixed.
This is not financial advice.
Source
Researched with AI assistance, fact-checked and edited by a human. Not financial advice.