AI complicates proof of life in the search for Savannah Guthrie's mother

+ Accessibility links Skip to main content Keyboard shortcuts for audio player Open Navigation Menu Newsletters NPR Shop Close Navigation Menu Home News Expand/collapse submenu for News National World Politics Business Health Science Climate Race Culture Expand/collapse submenu for Culture Books Movies Television Pop Culture Food Art & Design Performing Arts Life Kit Gaming Music Expand/collapse submenu for Music The Best Music of 2025 All Songs Considered Tiny Desk Music Features Live Sessions Podcasts & Shows Expand/collapse submenu for Podcasts & Shows Daily Morning Edition Weekend Edition Saturday Weekend Edition Sunday All Things Considered Fresh Air Up First Featured Embedded The NPR Politics Podcast Throughline Trump's Terms More Podcasts & Shows Search Newsletters NPR Shop The Best Music of 2025 All Songs Considered Tiny Desk Music Features Live Sessions About NPR Diversity Support Careers Press Ethics AI complicates proof of life in the search for Savannah Guthrie's mother The FBI says AI is complicating the idea of a proof of life message in the search for Savannah Guthrie's mom. Experts say AI needs just a few short clips of a person to render a convincing fake. National AI complicates proof of life in the search for Savannah Guthrie's mother February 9, 20265:13 PM ET Heard on All Things Considered

+

+ AI complicates proof of life in the search for Savannah Guthrie's mother Listen &middot; 4:03 4:03 Transcript Toggle more options Download Embed Embed "> <iframe src="https://www.npr.org/player/embed/nx-s1-5707415/nx-s1-9641691" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player"> Transcript

+

+ AILSA CHANG, HOST: We have been following the search for "Today Show" cohost Savannah Guthrie's 84-year-old mother, Nancy Guthrie. That search is now in its second week. The family says that they received a message from people who claim to have taken her from her home, but the family also acknowledged how difficult AI has made it to determine real proof of life. Here's Savannah Guthrie in a recent Instagram post. (SOUNDBITE OF ARCHIVED RECORDING) SAVANNAH GUTHRIE: We need to know without a doubt that she is alive. CHANG: We're joined now by NPR's Kat Lonsdorf to talk about all of this. Hi, Kat. KAT LONSDORF, BYLINE: Hey there. CHANG: OK. So just catch us up here. Where do things currently stand in this case now? LONSDORF: Yeah. Nancy Guthrie has been missing for more than a week after what appeared to be a routine night at her home where she lives alone in Arizona. Since then, the Guthrie family have posted several videos on Instagram pleading for her return. Meanwhile, there have been multiple reports of ransom notes being sent to various media, and the Guthries have said they are willing to pay a ransom. One TV station said they received a ransom note that demanded $6 million by sometime this evening. Police have not publicly confirmed details about that. But this concern about a fake proof-of-life video is something the FBI has also raised in this case, saying that AI could be used to create a convincing but fake video of Nancy. CHANG: Absolutely. And explain how realistic AI can be when it comes to fabricating a proof-of-life video. LONSDORF: Yeah. I mean, Ailsa, this technology has gotten better by leaps and bounds in the past year, even in the past few months, and it will only get better. One reason is because the barrier to being able to create a fake video has gotten so much lower. It doesn't take that much sophisticated technology or knowledge anymore. Almost anyone on the internet can do it. The other is that it used to take a lot of content to train AI to make a realistic video of someone - you know, hours of video or hours of recordings of their voice. So it was mainly a concern for celebrities. Now you really only need a few images and a few seconds of someone's voice. Here's how Hany Farid put it to me. He's a professor at the University of California Berkeley, who specializes in digital media forensics. HANY FARID: Even if you have a single LinkedIn profile and somebody listened to your voicemail... (SOUNDBITE OF FINGER SNAP) FARID: ...That's it. I have your identity. That's what it takes today. CHANG: God. LONSDORF: So someone - you take someone like Nancy Guthrie, who's been on the "Today Show" multiple times... CHANG: Yeah. LONSDORF: ...That's plenty of content for a bad actor or a scammer to make a relatively believable fake proof-of-life video. CHANG: I mean, are there effective ways to tell if a video is fake? LONSDORF: Yeah. I mean, it can be pretty convincing, especially to an untrained eye. It used to be that someone holding up a copy of today's newspaper and talking straight to a camera would be enough. Unfortunately, those days are gone. That kind of shot can be created fairly easily now. There are a few things that could make it easier to tell, though. You know, the longer a video is, the more opportunity there is for AI to make a mistake - you know, something being off in the background or an extra finger on a hand for even just a few frames - that kind of thing. Or a video shot from multiple angles is also more believable. That's much harder for AI to do for now. CHANG: At least for now. But I mean, Kat, if all you need is a photo and just - I don't know - a few seconds of someone's voice to make a relatively... LONSDORF: Yeah. CHANG: ...Believable video... LONSDORF: Yeah. CHANG: ...There must be a lot of opportunities for scams, right? Like, beyond a... LONSDORF: Yeah. CHANG: ...Proof-of-life video in a kidnapping. LONSDORF: I mean, exactly. And I think that's something that people need to be aware of. I mean, hopefully, not many of us will find ourselves in this awful situation that the Guthrie family is in. But fake videos and phone calls that sound or look a lot like loved ones will almost certainly become more prevalent in relatively simple scams - you know, asking for money, that kind of thing. Farid from the University of California Berkeley told me that one simple way to combat that is to have a passcode or a phrase you share with your loved ones that they can drop into a message if they're ever in distress to let you know it really is them. That's a very analog way to combat a high-tech problem, but it works, and it can help you distinguish a real message from a fake one in a short amount of time. CHANG: The things that we have to think about now. LONSDORF: Yeah. CHANG: That is NPR's Kat Lonsdorf. Thank you so much, Kat. LONSDORF: Thank you. Copyright &copy; 2026 NPR. All rights reserved. Visit our website terms of use and permissions pages at www.npr.org for further information. Accuracy and availability of NPR transcripts may vary. Transcript text may be revised to correct errors or match updates to audio. Audio on npr.org may be edited after its original broadcast or publication. The authoritative record of NPR’s programming is the audio record. Facebook Flipboard Email Read & Listen Home News Culture Music Podcasts & Shows Connect Newsletters Facebook Instagram Press Public Editor Corrections Transcripts Contact & Help About NPR Overview Diversity NPR Network Accessibility Ethics Finances Get Involved Support Public Radio Sponsor NPR NPR Careers NPR Shop NPR Extra Terms of Use Privacy Your Privacy Choices Text Only Sponsor Message Sponsor MessageBecome

+ (function () { var loadPageJs = function () { (window.webpackJsonp=window.webpackJsonp||[]).push([[22],{1167:function(e,n,c){e.exports=c(323)},323:function(e,n,c){"use strict";c.p=NPR.serverVars.webpackPublicPath,Promise.all([c.e(1),c.e(2),c.e(3),c.e(4),c.e(84)]).then(function(e){c(3),c(1140),c(116),c(94),c(52),c(493),c(239),c(102),c(104),c(1141),c(143),c(1142),c(238),c(48),c(1143)}.bind(null,c)).catch(c.oe)}},[[1167,0]]]); }; if (document.readyState === 'complete') { loadPageJs(); } else { window.addEventListener('load', function load() { window.removeEventListener('load', load, false); loadPageJs(); }); } })();