CyberWire Daily - AI, misinformation, and the future of cybersecurity. [T-Minus: Space-Cyber Briefing]

Episode Date: August 16, 2026

As AI products proliferate, they continue to introduce new concerns, which have subtly eroded trust in imagery and content created by space-based infrastructure. In this week's episode, host Maria Varmazis sits down with ⁠⁠⁠Dave Bittner and Brandon Karpf to look at Google's troubled implementation of Nano Banana 2 in Google Earth. The incident raises larger concerns regarding how AI systems are becoming deeper ingrained into everyday life despite their ability to be misused and spread misinformation. Like what you heard? Be sure to subscribe to our free Signals and Space Briefing⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠, our Sunday newsletter covering the intersection of cybersecurity and space. Subscribe at: ⁠⁠https://thecyberwire.com/newsletters/signals-and-space⁠⁠ Is there a topic or person you’d like to hear on our show? You can send your questions and feedback to ⁠⁠[e-mail]⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠. You can also fill our our audience survey: ⁠⁠https://www.surveymonkey.com/r/NJYCN2P ⁠⁠ T-Minus: Space-Cyber Briefing is a production of N2K CyberWire. ⁠⁠N2K⁠⁠ is your nexus for discovery and connection for people, technology, and ideas shaping the future of secure innovation. Learn how at ⁠⁠n2k.com⁠⁠.

You're listening to the Cyberwire Network, powered by N2K. Social engineering attacks look trustworthy, a routine request, an internal email, a familiar face on a call. But Dopple sees through the disguise. Their AI-native platform detects and disrupts attacks across every channel, trains employees to recognize deepfakes and deception, and investigates every fish to take down the campaign behind it. They fight relentlessly to protect you. your business, brand, and people.

Dopple, outpacing what's next in social engineering. Learn more at doppel.com. That's do p-p-e-l.com. When you lose that kind of small community of interest where the major players know each other, they talk, they share information, they share applications, they share use cases and ideas, right? That's the foundation of trust.

And the moment you kind of expand an idea and create a generalized concept or a generalized tool, you lose that community of interest and therefore kind of that root of trust. And that seems to be what we're struggling with across industries right now, especially across the internet and artificial intelligence systems and kind of what is the root of trust. And at the end of the day, it's based on relationships. And it's very hard to have a relationship, you know, across massive communities of people. Welcome. I'm Maria Vermazes. And you're listening to T-minus space cyber briefing.

In this show, we examine the evolution of cybersecurity in the global and orbital infrastructure that powers, protects, and connects our lives. Hello and welcome, everyone. It is always a joy to have a group chat with two of my favorite people, and they would be Brandon Karp, leader of international public-private partnerships at NTT, and my N2K colleague and host of the Cyberwire Daily, Dave Bittner. For our session today, we decided to take a closer look at the Goose. Google Earth, nano-banana debacle from late July.

No worries if you do not know what I'm talking about. In fact, we'll dive in first with Dave asking me to put my explaining cap on before we break it all down. Here we go. So, Maria, I'm going to ask you to do the introduction here on our topic. There was some brief fun in the world of Google Maps recently. What happened? Oh, goodness, was this fun? All right.

So on July 30th, to much fanfare across the entire internet, Google Earth announced that you could generate with AI, Nanobanana 2 specifically. Basically anything you wanted to do to a real-life place in Google Earth, if you can imagine it, you could do it. And the pitch from Google for this was amazing. They rolled out a fancy blog post for it and everything, and they had five justifications for why putting AI

and image generation into Google Earth was a great idea. Yes, because nothing says geospatial intelligence, like generative AI images. For the entire public, unfettered to use. So, yeah, so reason number one was that you could bring history to life. So if you ever want to do imagine what Pompeii looked like pre, you know, you could do that. While Vasubius.

While Vasubius was exploding. You could just do that. Watch people running in horror for their lives. Yeah, you can then. and walk through Pompeii, you know, as it was potentially. Okay. Number two was learn more while you explore, so you could create custom infographics about a place without leaving Google Earth.

Ah, yes. That is certainly what everyone is doing. Oh, yeah, definitely. Pitch number three was you could create professional real estate plans. So, you know, here's a blank spot of ground. Generate what your strip mall would look like in real time, I guess. All right. Number four.

That one makes sense. reason number four sort of seemed like reason number three. I guess they just needed to get to an odd number so they patted it a little bit was visualize projects before breaking ground. That sounds a little similar

that recent number three. Yeah. And then number five, give your favorite place of makeover. Oh, perfect. And boy, did people do just that. And the entire Earth observation community cried out in unison

and then fell silent. It was quite amazing. It was like millions of people. voices suddenly cried out. Now, on the brief period of time that this was up, and we'll get to that fact in a minute, but my favorite, I think, that I saw was Japan being attacked, Tokyo specifically being attacked by Godzilla. Now, Brandon, you were recently in Tokyo.

In fact, I was in Tokyo on that day. On that day! As our man on the ground, can you confirm to us that this was AI and Tokyo did not, in fact, victim once again to Godzilla? You know, the fine people of Tokyo whom I work closely with in my job at NTT, they do love

Godzilla, and there are numerous Godzilla statues all over that city. I, you know, I looked, I tried to find this Google Earth, you know, life-size Godzilla, and I couldn't find it in person, so

it must not have been true. Okay. It says shame. Yeah. So, Maria, the fund didn't last long. No, it did not. So again, Earth observation folks really did not like this, and it wasn't just cynically because it potentially take away money from them. It's because this is going to create a nightmare, which I'm sure we'll get into.

This feature from Google Earth lasted, I think, less than a day. Pretty much right after the launch, I think several hours later, Google Earth actually pulled it, and then they updated the blog post where they announced this feature the next day on the 31st of July saying, we know that people uniquely trust Google Earth for a reliable view of the world. We've seen geospatial professionals using this feature for a range of useful purposes. However, we've also seen people sharing screenshots of generated imagery that appear to violate our policies. So we're rolling back this feature in Google Earth while we work on implementing stronger guardrails.

It's important to note that generated images didn't appear in the main Google Earth experience for others to see and were watermarked as AI generated. That was their update. Okay. So I'm sure that would have stopped any shenanigans. This message brought to you by Google's legal team. Yeah. So, I mean, this whole thing, it's very funny. And I love the Internet.

And it's delightful when the Internet reminds you why it was invented and the sharing of memes and Cat videos and Godzilla attacking Tokyo through Google Earth. but it kind of brings up this point that we're still talking about across multiple industries, which is what are the legitimate useful applications of AI in all of these industries? And I think the space industry in particular has been struggling with this one quite a bit and hasn't really figured out the useful applications of some of these technologies. And so experimenting is good.

Experimenting is useful. but I think that there's a lot to still be figured out by people in the industry. Yeah. I'm not sure what the right application is for Earth observation, but generating fake images,

certainly isn't it? Yeah. Yeah, I just, what could possibly go wrong is just the question everybody should ask with this sort of thing, and I'm amazed that this made it to print,

so to speak. I just... Does this set a new record for a Google project being removed. I mean, Google sort of famously deletes things that people have found useful, but I don't think

it's ever happened this quickly. Yeah, within 20 hours. It kind of reminds me of that Microsoft chatbot that they put on Twitter, you know, a decade ago. Yeah, TAY. Yes. That became instantly racist.

Yes. Oh, my God. Yes. I forgot about that. Oh, yeah. My take on TAY was that TAY reflected who we actually are, not who we aspire to be. And that was Tay's downfall.

Who we are, at least on the internet. Right. The sum total of all good and bad, of all earth's people. Yeah. So at the risk of taking this more seriously than perhaps it deserves to be taken, what are the takeaway lessons here when it comes to, shall I say, slapping AI on everything that moves? Well, I think the,

At a certain extent, some of these companies have to slap AI on everything that moves to justify the capital expenditures and to experiment in such a way to find a useful application. So it's striking a difficult balance, and this is kind of the security angle as well, is these industries are investing so much capital and human resources and time that they need to see where this can apply and be useful. And like that one use case that you brought up, Marie, in the beginning of, okay, let's test out maybe some, you know, large construction projects and see what it would look like and how it would affect other regions and the shadows it would cast or how it would affect traffic flow and things like that. All of that actually sounds like a fantastic use of AI, you know, connected to some, you know, legitimate applications and products. But the way to make that useful is putting guardrails around it and actually can kind of very deliberately kind of making the use case as specific as possible. I think the problem here is it was as broad as possible. There were no guardrails.

It was a general addition of image generation to this tool that is kind of like doesn't have a specific use case. It has whatever. It's a creative system. Yeah. People can do whatever they want with it. And some of those use cases will be helped. helpful like real estate projects and others will be silly like, you know,

uh, kajus attacking cities and some of them will be downright disturbing, like potentially air, you know, images of air strikes in, uh, in Warsaw or something like that. Yeah.

Yeah, which a lot of those images were, uh, just made for demonstration purposes of this is how easy it is to show disinformation. Uh, and, and,

and just, I saw it everywhere on social media going, this was real easy to fabric, fabricate an entire story now with a convincing Google Earth image. Yeah. Yeah, I mean, the use case of, again, a real estate project or something

like that, I mean, that's been around for a little while but privately, just the idea of making this publicly available. I mean, it's sort of one of those I struggle with it because, yeah, information wants to be free. Why shouldn't everyone get to play around with this kind of thing? But maybe we shouldn't. I don't know.

I struggle with it. But the disinformation potential for this was so obvious. I'm just amazed. And the idea of, oh, we put a watermark on it, that has not stopped believing all these images that are out there and they're easy to remove anyway, to be honest. Please don't ask me how I know how easily one can remove those watermarks. I really don't want to get in trouble.

We'll be right back. Let's get back to my conversation now with Brandon Karp and Dave Bittner about the Google Earth AI image generation, shall we say, missed up, Dave will start us off. I guess for me, the, and this is part of a bigger concern I have with AI in general, is just the potential for the erosion of

ground truth. Yeah. Yes, absolutely. Google Earth is one of the pillars on which we build ground truth understanding of our world. And so to add a tool that makes it easy to manipulate seems problematic to me.

to completely undermine its trustworthiness is why would you want to tank your brand that way? That's a whole marketing discussion, but that's just, I just, I don't understand it just completely. Yeah, I mean, in terms of in the space context, Brandon, going back to a much earlier point you made about applications for AI, the one that I'm really excited about is for autonomy, for spacecraft, where there's, you know, they're really far away. That makes a lot of sense to me. I mean, we're not there yet. Yeah, we're just in the beginning stages. Yeah. We're seeing like Starfish space, you know, autonomously doing rendezvous in orbit, right?

So like those control systems are running and some of that is, I imagine, reinforcement learning or some sort of what we would consider to be artificial intelligence, right? Some sort of algorithmic type of intelligence. And so, yeah, to your point, like there are some really cool use cases. And then in terms of Earth observation and applications on the ground and how we make use of that data or process data in situ in space before sending it down. There's a lot of really interesting, but again, very niche, extraordinarily niche applications of this stuff. And today's point, when you lose that kind of small community of interest where the major players know each other, they talk, they share information, they share applications, they share use cases and ideas, right, that's the foundation of trust.

And the moment you kind of expand an idea and create a generalized concept or a generalized tool, you lose that community of interest and therefore kind of that root of trust. And that seems to be what we're struggling with across industries right now, especially across the Internet and artificial intelligence systems and kind of what is the root of trust. And at the end of the day, it's based on relationships. And it's very hard to have a relationship across massive communities of people. Yeah, I have to wonder if, Maxar had Google Earth on speed dial

and just the moment this was announced, they said, bro, you cannot do this. I just have to wonder that industry pressure at play. I mean, it's going to cast out on everything, right? Yeah, exactly. And like another implication of that. I just saw, I wish I remember where I saw it, but just in the last couple days,

this fascinating demonstration how Google Earth images are being updated in Ukraine. And so you're having now older pre-war Ukraine images stitched right next to post-war images. And so these towns look like they're like half destroyed with a very hard transition line because some of the images were taken five years ago and some of them were taken in the last couple years. And so you're seeing in real time what's actually the before and after, which I think is actually really useful to see in a really interesting kind of secondary impact.

of these technologies. But then, to your point, right, if they implement a technology on top of Google Earth that calls into question the veracity of those images, that doesn't help, you know, the norms of society within which I know we all want to live. Yeah, I mean, it would be nice. People just shrugging and saying, ah, AI, is that the new fake news? You know, it just makes it easy to dismiss anything. Yeah. Yeah, I think we're already there.

Yeah, I think we're already there. I mean, I'm already there with a lot of this stuff. I don't know. And this felt like another domino falling. It just didn't feel good. I once pranked my nephew by convincing him that Google Maps, the satellite views were real time. So I had him.

A lot of people think that, though. I had him run outside. Yeah, I had him run outside the house and wave. Then he came back in. I said, I totally saw you waving out there. So I'll leave you with a similar funny story. In college, there's this buddy of mine at the Naval Academy. And when we'd go out to bars in Washington and D.C.,

we didn't necessarily want to say that we were midshipment at the Naval Academy. And so he came up with this great idea where he told people when he was meeting them, typically other young women, that he was part of the Google Sky Team. And the Google Sky Team purpose was to, skydive and take a camera with them and videotape the skydive. And that's how you got the Zoom on Google Maps and Google Earth. And the number of people who believed this was extraordinary. Honestly, a couple beers in, I'd believe it. I love it. I'm not even going to lie. If he sold it well, I'd be like completely on board. I want to believe that that's true. Sam, Sam, wherever you are,

you sold it very well. Oh my goodness. Wow. All right. Well, Brandon Carf is leader of international public-private partnerships at NTT. Maria Vermazas is my CyberWire co-host and host of the T-minus Space Cyber Podcast. Thank you both for joining us. Dave and Maria. Great to see you all. Good to see you, Brandon. Good to see you, Dave. And that's T-minus Space Cyber Briefing, brought to you by N2K Cyberwire. If you like what you heard today, you will also enjoy our newsletter, Signals in Space.

You'll get research and notes pulled together by our producer Ethan Cook and me. along with this week's top space cyber news stories. Subscribe by visiting thecyberwire.com slash newsletters. And as always, we would love to know what you think of our podcast. Your feedback ensures we deliver the insights that keep you a step ahead in the rapidly changing cybersecurity landscape. If you like the show, please share a rating and review in your podcast app. These also fill up a survey in the show notes or send an email to space at n2K.com.

We're proud that N2K Cyberwire is part of the daily routine of the most influential leaders and operators in the public and private sector, from the Fortune 500 to many of the world's eminent intelligence and law enforcement agencies. N2K helps cybersecurity professionals grow, learn, and stay informed. As the nexus for discovery and connection, we bring you the people, the technology,

and the ideas shaping the future of secure innovation. Learn how at N2K.com. Thank you for listening to T-Minus. I am your host, Maria Vermazas. And the show is produced by Ethan Cook and Liz Stokes. We're mixed by Elliot Peltz, and Trey Hester, with original music by Elliot Peltzman. Our executive producer is Jennifer Ibin, with content strategy by My On Plout.

Peter Kilphee is our publisher. We'll see you next week.