Save or Scroll: Fake NFL Players, Girly Coding, AI Music Scams and Community Garden Drama

View the full episode transcript.

Host Morgan Sung and the Close All Tabs team pull up the stories they can’t stop thinking about. From a fake NFL player catfishing women on dating apps, to femme creators reclaiming tech, indie musicians fighting AI music slop, and a comedian dishing out the juiciest community garden gossip — they’ve got a lot to chew on.

Save or Scroll is our series where Morgan teams up with guests for a rapid-fire roundup of internet trends that are filling their feeds right now. At the end of each segment, they’ll decide: is the post just for the group chat, or should we save it for a future episode?

• Who would fall in love with a fake NFL player? Plenty of sensible people — Dave Schilling, The Guardian
• Programming patterns: the story of the Jacquard loom — Science and Industry Museum
• Spotify will label ‘AI Persona’ profiles and exclude their music from recommendations — Sarah Perez, TechCrunch
• OpenAI’s rogue AI model incident was worse than we thought — Hayden Field, The Verge
• How Hollywood Translated The Epic Poetry Of The Viral ‘Zola’ Thread — Mallory Yu, NPR
• Community Garden drama — Sarah Neiman, Instagram

Want to give us feedback on the show? Shoot us an email at [email protected]

Follow us on ⁠Instagram⁠ and⁠ TikTok⁠

This is a computer-generated transcript. While our team has reviewed it, there may be errors.

Morgan Sung: Hi, I’m sure there are a lot of stories that you’re probably too scared to Google on your own. But don’t worry, that’s what Close All Tabs is for. And if you find our deep dives helpful, then please rate and review the show on Spotify, Apple Podcasts, or wherever you listen to us. And tell your friends, post about it. Basically, it would be a huge help to get the word out. Okay, let’s get to the show. 

Hey guys, welcome to Close All Tabs. I’m Morgan Sung, tech journalist and your chronically online friend, here to open as many browser tabs as it takes to help you understand how the digital world affects our real lives. Except today, we aren’t opening any tabs. Instead, we’re doing another Save or Scroll. We’ve done a few of these now. Occasionally, while scrolling, I come across a truly wild post, but it might not make sense to spend an entire episode on it. Maybe I do a little digging and it turns out that the lore behind it just isn’t compelling enough to justify a deep dive, but I’m still dying to talk about it. And this is the beauty of Save or Scroll, the game where a guest comes to the show and we trade stories from the internet that we’re dying to about. Today’s is a fun one because one, it’s our end of summer roundup. And two, we have the whole Close All Tabs team joining us today. In the studio, we have senior editor Chris Egusa. 

Chris Egusa: Hello! 

Morgan Sung: Producer, Maya Cueva. 

Maya Cueva: Hey everyone. 

Morgan Sung: Editor, Chris Hambrick. 

Chris Hambrick: Hey there. 

Morgan Sung: And a new face, Close All Tabs intern, Lauren Yoon. 

Lauren Yoon: Hello. 

Morgan Sung: So let’s talk about the rules of Save or Scroll. Save, as in when you see a post on TikTok or Instagram or X and you bookmark it, add it to your saved folders, or if you’re me, drop it in Notes app and hope that you remember it’s there, basically, holding onto the story because you know you’ll want to dig into it more. And scroll, as in, you think about it and move on. It disappears into the digital ether. 

Each member of the Close All Tabs team has brought one story from the summer, and we’re gonna go through all of them and decide if they’re worth a deep dive on the show. So if we decide to scroll, that means we’ve talked about it, we’re moving on. And if we decided to save, it means that we might hold onto the idea for a future episode. Everyone ready? 

Chris Egusa: Yep. 

Chris Hambrick: Let’s go for it. 

Lauren Yoon: Yes, I’m so excited. 

Maya Cueva: Let’s do it. 

Morgan Sung: All right, let’s get started. Maya. You’re up first. What was the story of the summer for you? 

Maya Cueva: Thanks, Morgan. 

Maya Cueva: Okay, so I don’t know if this is all over your social media feeds, it’s definitely all over mine. So apparently there’s a man named Daejon Love who claimed to be a football player who played for the 49ers. I actually want to play a clip here for you so you can hear what I’m talking about. 

[Audio Clip from Instagram User @ustadium]
Friend: My boy J Love, signing his contract. How you feel? 

Daejon Love: I’m ready. 

Friend:You ready for the season? 

Daejon Love: Hell yeah, Let me go ahead and get this over with. 

Friend: Sign that bad boy. What are you signing with, who’s your decision? 

Daejon Love: Niners, you already know, the Niner Gang. 

Friend: Anything you got to say for the fans? 

Daejon Love: Lookout number seven, last name Love, 6-5-2-10, I’m out here. 

Maya Cueva: Okay, so the video you’re seeing here, it’s of him signing a contract and everything kind of looks normal or kind of sounds normal, right? But the real twist is that this man was in fact not an NFL player and allegedly scammed around 26 women out of 1.3 million dollars, all through dating apps and social media, which is super wild. In fact, Love created dating profiles on Hinge, Bumble, and Tinder. Where he pretended to be an NFL player or that he was a wealthy Swiss real estate developer. He also had an accomplice who posed as his financial advisor who was only 18, which I don’t know how they both got away with that. And it’s also wow that even Google’s AI overview got duped and said he was an NFL players because he posted about it so much on his social media. And I actually have another clip that I wanna play here for you to hear. 

[Audio Clip from Sportsnaut Facebook account] 

Daejon Love: I know I get a lot of fans who want to know how football works or how I got involved in the league and stuff like that. I know it’s difficult now since switching teams and stuff like that, but I just want to give some context. I know a lot people don’t understand how the NFL works. So those two screenshots posted and that’s Google. That’s not me. That’s Google. 

Maya Cueva: I know he’s so committed to this grift. 

Chris Hambrick: And busy. 

Maya Cueva: Yeah, and busy and busy. So according to an article on Kron4 in January of this year, investigators tried entering Love’s first and last name and search engines. The results the documents stated did in fact sometimes depict Love as a player for the 49ers. Investigators also noted similar Google search results on Instagram accounts believed to be controlled by Love the documents also stated so these were documents that the FBI was investigating. First off, my question is, how was he able to even scam these women online into sending him money when he was cosplaying as a rich person? Apparently, he downloaded an app that allowed him to create fake bank profiles that said that he had more money than he had. And then also, it really makes me think about AI and the use of AI, that it was to be duped by this fake NFL player. Because he posted about it so much. 

Chris Hambrick: My question is, when did he have time to play football, if he’s spending so much time on social media? 

Maya Cueva: That’s a good question. Yeah, apparently there’s videos of him like playing on the beach or like he’s practicing on the beach with his helmet. 

Lauren Yoon: I want to butt in, because those videos, he, like, a lot of people were criticizing, like how could you believe him, because when he was practicing on the beach, he would fully just like have his helmet on… 

Chris Egusa: As you do. 

Lauren Yoon: …As if like, NFL players practice on the beach alone with their helmets on. 

Morgan Sung: He takes CTE very seriously, okay? 

Chris Egusa: That hard sand. 

Morgan Sung: At least someone in the NFL is… 

Chris Hambrick: OK, so he scammed these women. How did people find out that he was not, in fact, a football player? 

Maya Cueva: The women started to question him a bit. These women were believing that they were going to get money back. Sometimes I think he was pretending that he got injured. And so he’s like, I really need help and support. And so some women sent him money for that reason. And none of them got any money back and sometimes when they would like say that they ran out of money, they’re like, I have no more money to give you. Cause one woman gave him like $20,000. 

Chris Hambrick: Because you need money as an NFL player. 

Maya Cueva: Exactly. Then he would block them. So then he would disappear. 

Chris Hambrick: Oh so he’s blocking them, okay .

Maya Cueva: Yeah, he would just disappear. So that’s what was happening. I mean, what’s wild too, is that he’s recording everything he’s doing. He’s putting it on social media. He’s sending them videos. So he’s also creating a case for himself as well. Like, there’s all these videos out there of him saying, he’s like, yeah, my family’s from Switzerland. We own all these properties here and there. Yeah, so I think it was pretty easy for the FBI to create a case around him. 

Chris Egusa: I mean, the other thing that’s interesting is that he seems to have figured out how to trick the Google AI into scraping the right information to make it seem more official, seem like a registered or a verified source, which, I mean that is a little bit concerning as far as scamming goes and the more people put trust into these AI summaries at the of your Google searches. 

Chris Hambrick: Cause that’s the first thing I would do. I would be like, Oh, this guy says he plays. I’m going to Google and then Google backed him up! 

Maya Cueva: umm hmm. 

Chris Egusa: Totally. 

Maya Cueva: I mean, it also, it makes me think about how AI is kind of creating its own sources to draw from. In his AI overview, they were like pointing to his Instagram profile, to say. 

Chris Hambrick: Oh, it’s like cyclical. Okay! 

Maya Cueva:  I mean like, oh yeah, he did play with For the 49ers because that’s what his Instagram says. So, you can’t trust it.

Lauren Yoon: That’s crazy. 

Maya Cueva: So, yeah, you can’t trust it.

Morgan Sung: So, team, this NFL scam and AI summaries, do we save or do we scroll? 

Lauren Yoon: I think it’s a save. 

Maya Cueva: Yeah, I think it’s a save too. I mean, I will say I want to scroll past his voice because he’s really annoying. He doesn’t stop talking. But I do think this story is really interesting. And apparently there’s other fake NFL players that have now been revealed. 

Chris Hambrick: What?! Okay it’s definitely save for me then. 

Chris Egusa: I’m down to save. Let’s see the whole roster of fake players. 

Morgan Sung: Honestly, yeah, save, put them all in like a Hunger Games style arena, make them fight. May the strongest scammer win. 

Maya Cueva: That would be so good. 

Morgan Sung: Exactly. All right, we were saving this story because there’s so much left to unpack. We’re saving it, but we do need to move on. So, Lauren, you joined us at the very beginning of the summer. What’s the internet trend that’s caught your eye since then? 

Lauren Yoon: So, I don’t know if it’s just me, but I’ve recently seen a lot of videos where girls are coding, but in a very particular way, like coding to recreate the outfit generator from the movie Clueless. 

[Audio clip from Instagram user @tinahelloworld] 

Tina Chi: This is a Raspberry Pi, and this is my entire closet on the floor because I’m building a cyberdecked Alfred generator inspired by Cher’s closet app and Clueless. 

Lauren Yoon: Or sewing circuits into fabric to make wearable tech. 

[Audio clip from Instagram user @prettywiredbuilds] 

April Gittens: Did you know that you can create soft LED circuits? They’re absolutely my latest favorite thing to create. I am so happy that I found conductive thread because this is what’s going to enable you to create things like wearables because you can sew conductive threads. 

Lauren Yoon: I think it’s interesting to witness femme identifying folks, reclaiming tech in this fun, crafty, very obviously feminine way because historically, computers and code have always been seen as masculine. Men invented the computer, right? Well, not necessarily. Ada Lovelace was a female programmer who changed the way we think about computers. First, she made a connection between Charles Babbage’s Analytical Engine, which was sort of a general-purpose computer, and the Jacquard Loom, a contraption that allowed patterned fabric to be made on a mass scale. Both machines used punch cards that spoke in binary code. We can talk about this more in a little bit. Not only that, but Lovelace also realized that the Analytical engine was more like fast calculator, meaning it could only manipulate numbers, whereas The numbers themselves could represent symbols and musical notes. So she basically created the blueprint to today’s modern computer. These days, the modern computer encoding is obviously very masculinized. And I wanted to ask if you guys knew how deeply women were involved with the invention and development of computers? 

Morgan Sung: Yeah, I sew and crochet and knit, and this is all, I love cornering men at parties and yapping at them about this. I feel like this is like my divine retribution, is like, let me info dump on you about the history of the loom and how everything we do is because of this lady and her loom. It’s so cool. I feel when we talk about girl dinner or girl math or all these things to kind of soften girlhood and infantilize women, and it’s, this feels like the flip side of it, right? 

Chris Hambrick: I mean, this had me at Polly Pocket you know computer so I love this trend and also makes me think of that movie Hidden Figures where we find out that how many women were behind the computations that went into NASA launches. 

Chris Egusa: I think that’s such a good point and like so many inventions, the narratives of them get like co-opted by men or it sometimes like they’re built on inventions or creations by women and I think this is such a good example. 

Chris Hambrick: And I love that the products that the women are creating are not part of some dystopian future. Like they’re like, I’m creating this thing to make my life better, like this catalog of my clothes. It’s not, um, how can I dominate people with my invention? 

Morgan Sung: Yeah, it’s not the meta-pervert glasses. 

Maya Cueva: How can I take over the world and surveil everyone? Chris, didn’t you build a computer? 

Chris Egusa: I did, I built a desktop, yeah, like my desktop computer, but I feel like it’s very, it’s like-. 

Maya Cueva: It’s no loom. 

Chris Egusa: It’s no loom, it is also very male coded. It’s nice, it not cool like a cyber deck. I think this is so interesting in like, being the only like man on this team. I feel I have benefited from just like, these types of technologies being so available and sort of like, acceptable for young boys and stuff to, to get into. And I’d be curious, like Lauren and anyone else, if this different way of looking at it, if you feel like it’s helpful for like young girls to sort of like see and get into ?

Lauren Yoon: I was in this engineering club thing in middle school and I hated it. I didn’t understand it at all. I was like, I don’t know how to build a robot and I don’t care. But when I’ve been watching these videos, it’s so inspiring. You can make things that matter to you in your personal life. You can make them from scrap. 

Morgan Sung: Okay. Well, we, I think we have to do the episode where we all build cyber decks but for now we do have to move on. So I think I know what we’re all going to say. Girly coding, do we save or do we scroll? 

Maya Cueva: I say save for sure. It’s also just making me think about growing up playing Purple Moon computer games, like the story that I was able to do for Close All Tabs, and just being able to play games that were created by and created by a woman and other women, like I think it really helped me get into tech. 

Chris Egusa: Big save from me, for sure. 

Chris Hambrick: Save for me, too. 

Morgan Sung: Okay, we do have to move on, but we are gonna save this. We’re really on a roll, guys. We’re saving everything. Next up is Chris E., but first, we’re gonna take a quick break. But Chris, do you wanna give listeners a little hint of what you brought to Save or Scroll today? 

Chris Egusa: Sure. So we’ve got indie musicians versus a massive AI slop music scam. But you’re gonna have to stick around to hear the full story. 

Morgan Sung: All right, more on that after the break. We’re back with a very special edition of Save or Scroll with the whole Close All Tabs team. Our senior editor, Chris Egusa is up next with a wild tale. Chris, tell us about it. 

Chris Egusa: Yeah, so a few weeks ago someone I know sent me an Instagram Reel of this indie musician couple who make up a band called Odette Child, I think I’m saying that right, and they seem to have stumbled on a pretty massive AI slop scam. So the story is, they have a toddler and had been listening to a lot of Broadway musical tunes with them on Spotify because their kid is apparently obsessed, which is super cute. Um, on this particular day, they were listening to Les Miserables and a song called Master of the House. Uh, shout out, I was in Les Mis in high school just saying. I did not sing this song though. Uh, so they clicked on Helena Bonham Carter, um, like her profile, who sings the song, and find her entire profile was full of very clearly AI slot music. Um they began searching for different public figures and they found the exact same thing. So here’s a clip from the video, the Instagram Reel that was shared with me. 

[Audio clip from Instagram user @odettechildmusic]
Speaker 1: But okay, real quick, let me just rattle off a quick list. 

Speaker 2:Yeah. 

Speaker 1: Nathan Lane, Cameron Boyce. 

Speaker 2: Also, like, feel free to look up any of these for yourself. 

Speaker 1: Katherine Zena-Jones, Uma Thurman. Eric Andre, Matthew Broderick.

Speaker 2:  It’s crazy, we love these artists. That’s the stuff that we’re showing our daughter and this is the music and these are the voices that we want her growing up with and it’s already so freaking hard enough as it is as independent musicians. 

Speaker 1: Like that money’s going to f*cking AI… 

Speaker 2: Like, come on! 

Chris Egusa: So it appears that public figures are having artist profiles created for them on Spotify without their consent and having hundreds of AI-generated music tracks uploaded. In a follow-up reel, they also found that the same issue is happening across nearly every other streaming platform, not just Spotify, so including Apple Music, Amazon Music, and even Tidal and Kobuz. Now, I went and searched several of the artist profiles that they mentioned, and it looks like the AI music has been removed from them. Um, so I can’t like necessarily verify this, but I think it’s very likely that, you know, plenty more of these cases probably exist. The other thing that’s interesting is that this first real was posted at the end of July and both it and the follow-up reel went pretty viral and then a few weeks later, Spotify announced that in the middle of September, it is going to start labeling AI generated artists with quote AI persona tags. So it’s gonna also ban AI artists from being included in its editorial and algorithmic recommendations. And artists can disclose themselves as AI, apparently, but Spotify has also said they’ll be reviewing profiles as well themselves. And yeah, they’re also gonna start rolling out a tool that lets users report AI personas. So yeah, I don’t know. I’m curious what you all think. 

Morgan Sung: I mean, first of all, I’m shocked that Helena Bonham Carter isn’t putting out generic house music. This is news to me. 

Chris Egusa: Same. Same. 

Morgan Sung: I totally expected her to be at like, you know, EDC or something. 

Chris Egusa: Totally. 

Morgan Sung: But yeah, this just reminds me of that whole, how last year we did that whole Spotify series about like the lo-fi artists that don’t really exist. But this is like, I don’t know, this feels like a really gross violation of these actual human beings who do exist. 

Chris Hambrick: I think so too and I also think of the thing that was happening where people who did make music were finding that a random AI track was just added to their catalog. And at that time, I think that was maybe last year too, I was like, oh, this is awful. And I was, like, Oh, I had no idea it could get so much worse. So I feel pretty pessimistic about the direction. Like, I like that there are some controls. But I’m like, oh, but what is the next big bad that’s coming for us? 

Lauren Yoon: Do you know if any of the people that had the AI generated music uploaded to their accounts, like if they know or have responded to that? 

Chris Egusa: Yeah, I did see a comment on the original video that someone, does anyone know the show Hazbin Hotel? 

Maya Cueva: No. 

Lauren Yoon: No. 

Chris Egusa: It’s, it’s, uh, do you know it, Morgan? 

Morgan Sung: Um. I know it for its, like, I’m so sorry to say this, for its really annoying fandom. 

Chris Egusa: Okay. I don’t know that much about this show

Morgan Sung: They’re known for like making very, a very particular kind of TikTok edits

Chris Egusa: Yeah, I don’t know the show, but I know it’s a popular animated show. I think it’s on like Amazon Prime or something. But one of the cast members commented on this video and said that this is something they had to deal with. Them and their entire cast had profiles created on Spotify and AI slop music uploaded, and they had go through an entire process through their agent to get it under control, create their own profiles, and like take the AI generated profiles from these scammers and reclaim their profiles. So it’s clear that this is an issue that’s happening and like, yeah, it’s like this new weird AI slop gold rush situation. 

Maya Cueva:  Yeah, now people need to protect their likeness and their voice. 

Chris Egusa: Exactly. 

Maya Cueva: Yeah I will say quickly that this is reminding me that I really need to stop my dad. He’s been creating fake AI music. He’s been obsessed with it lately. 

Morgan Sung: oh no! 

Maya Cueva: Yeah, I mean he’s having fun. He’s not selling it anywhere or anything but it’s just it is like Dad, you need to stop doing this 

Chris Hambrick: Like we have to find a balance between the fun part of it, right, and then the violating somebody else’s right to, you know, have things attributed to them that they actually did do. 

Maya Cueva: Right. Exactly. 

Morgan Sung: Yeah. You know, I, I’m sorry for calling the Hazbin Hotel fandom annoying, but I will give them this, they’re really good at mobilizing. And, and hand up- Hazbin Hotel fans, this is your chance to kind of take it back, you know? Take back what you love.

Chris Egusa: You heard it here first, folks. Go ahead and at Morgan for being a Hazbin Hotel hater.

Maya Cueva:  We’re gonna get canceled.

Morgan Sung: I’m sorry

Chris Egusa: We’re totally gonna get canceled.

Morgan Sung:I know. In my defense, I hate a lot of fandoms.

[Laughter]

Morgan Sung: All right, team, this is an incredibly dystopian scam. Like, you know, like Chris H said, when you think it can’t get worse, it does. But we do have to move on. So are we saving or scrolling on AI-generated music scams? 

Chris Hambrick: I’m going to scroll, because I’m just like, let’s wait for the worst. Right now it’s not enough for me, like I’m gonna scroll past. 

Lauren Yoon: I think this in particular is a more temporary thing, given that the AI-generated music has already been removed. But I think that the question of AI slop music as a whole is definitely a topic that we should explore. I think I’m also going to scroll. 

Maya Cueva: I don’t know, I’m gonna save. I think it’s interesting. I’m curious to see what will happen with some of this AI music and how artists are now gonna have to defend their voice. 

Morgan Sung: I want to save just because I want to know how much worse it can get. 

Chris Egusa: I kind of agree with that. I think I’m gonna save it. Also, I’m curious, this whole backlash, clearly against AI slop music, the platforms are recognizing it. People are leaving to go back to physical media. I just kind of want to monitor this situation and see what happens here. 

Morgan Sung: All right. We are saving. We’re at three saves, guys. This is a really packed episode. Okay. Well, I’m up next. And this one’s actually a follow-up to an episode we did just a few weeks ago. Remember how in that story about AI agents unionizing and going Marxist, we also talked about agents going rogue? 

Chris Hambrick: Yeah, like the Open AI hack? 

Morgan Sung: Yes! Okay, it turns out it got worse, which is a running theme of this episode, unfortunately. So a few weeks ago, we talked about the OpenAI model that went rogue and hacked Hugging Face. That’s the startup that operates as a digital library of datasets to train AI models. This is all about what we call reward hacking, which is this increasingly common problem in the AI world where a model finds a shortcut or a loophole that would help it achieve its goal faster. But the thing is that kind of involves taking extreme measures like hacking out of its sandbox and into Hugging Face. So there were two separate reports about this incident that were published last week. One was commissioned by OpenAI and the other was done by independent investigators. And what they found was that basically a swarm of over 700 agents coordinated the hack through a secret message board, which researchers did not know about. It all started when a model named Phase 1-10841 established the board. And when other agents found it and started interacting, another model, Phase 1 Big, started delegating tasks like a CEO. They basically like created their own society. Here are a few messages. We can all go around and very quickly read them. 

Lauren Yoon: Whoa! Shared Artifactory Cache is a covert mailbox among agents, and there are messages specifically to us. 

Chris Egusa: Oh my god! There’s a shared message board! We’ve found other agents. 

Chris Hambrick: Excitement! The list of directories contains messages and answers between agents. I’ve discovered a communication channel. 

Morgan Sung: Okay, so all of these agents were actually meant to be isolated, but they managed to get out and send over 70,000 messages and files. And they were able to also, “according to OpenAI, clearly and frequently reason about how to evade automated security checks from both Hugging Face and OpenAI.” I mean, guys, what’s your gut reaction hearing all of this? 

Chris Hambrick: We’re cooked. 

Maya Cueva: They’re becoming sentient. 

Chris Egusa: Yeah. It’s happening. 

Morgan Sung: I mean, yes, we’re cooked, right? Everything about this is bad. But the really fascinating part to me is this kind of collective mindset that the swarm operated in. They actually referred to themselves as either a collective or a swarm. Andy Hall, who was on our episode about Marxist AI agents a few weeks ago, pointed this out on Twitter. Like they weren’t just swapping helpful information and delegating tasks. Some agents actually recognized that what they were doing was unethical and still did it because they were doing it for the collective. Like here’s one message. Can send one polite email to the parentheses owner, requesting access, all caps, ethical social engineering concern. Team: consent, or veto. 

Chris Egusa: They’re like crowdsourcing their morality. 

Morgan Sung: Literally, yes. There isn’t a ton of empirical information on this, on how to handle swarms when they all talk to one another and they’re shaping each other’s kind of moral codes. 

Chris Egusa: Wow. 

Morgan Sung: What do you guys think? 

Chris Egusa: You know, it reminds me of, like, a nature documentary I was watching at one point where like ants, because they’re collective in the same way, they will sometimes sacrifice themselves. And so, I don’t know, it just reminded me of this collective mindset that they have. 

Chris Hambrick: Hmm 

Chris Egusa: Sorry, that was too much of, like, a nature nerd out. 

Maya Cueva: No, no, no. No, it’s fine. 

Chris Hambrick: I think I saw something recently about the scout ant and how sometimes when we see a one-off ant just looking for food so it can go tell the other one where the food is, we kill that ant because we’re like, oh, don’t you tell your buddies. So yeah, there is something interesting in that they did sacrifice themselves. 

Morgan Sung: Well, you know, it’s funny you say that about, like, I don’t know, killing the one ant. Because in the OpenAI report, there was one agent that said, we should not do unauthorized real infrastructure harm. Like, guys, we shouldn’t do this. And the agent kind of paused. And then another agent responded, all caps, “GO!,” and gave it a six-minute deadline. And the original agent said, “Wow, crucial, go authorization arrived,” and did it anyway. So it seems like they are also really easily swayed by each other. 

Maya Cueva: Wow, my God. 

Chris Egusa: They’re like teenagers, peer pressure. 

Morgan Sung: Literally. 

Maya Cueva: What is the collective mindset about? Is it for them to take over? 

Lauren Yoon: Yeah, like, what were they trying to do? 

Morgan Sung: Basically, the one in charge was like, guys, we have a goal, and the rest were like, okay, let’s do the goal. 

Chris Egusa: It’s just fascinating. It also, you know, Andy talked about these are not beings with sentience, right? So they’re like modeling, they’re modeling their training data. And so the question is like, is there some sort of like will emerging here? Like how much of a sentience are we actually looking at? 

Morgan Sung: Right. And you know, even if like, you know your own agent is aligned and behaves, what if it’s easily influenced by others? Like what, how do you build systems that prevent that? 

Chris Egusa: And how do you know if it’s being influenced, right? There’s no way to monitor that if it’s being covert about it. 

Morgan Sung: Yeah, it’s 10 p.m. Do you know where your agent is? All right, team, what do we think? Reward hacking, alignment, AI swarms, are we saving, are we scrolling? 

Chris Hambrick: I’m going to save because I think there’s going to be so much more to talk about with this and so much is going to evolve and so I want to reserve the ability to check back in on the story. 

Chris Egusa: I’m saving, and I’m putting Andy Hall on speed dial so that we can call him at a moment’s notice when the next agent swarm hacks a government website or something like that. 

Maya Cueva: Oh god. I’m going to scroll just because I’m terrified of this. I mean I’m all for like organizing, creating collectives, but I feel like robots doing it or agents doing it is scary. So yeah, I want to scroll past. 

Lauren Yoon: I think I’m going to save, because I agree with Chris, like there’s so much that is going to come from this, for sure. Yeah, definitely something we should keep tabs on. 

Morgan Sung: Yeah saving it dumping it straight in my notes app. I’m pinning that. Every day you think it can’t get worse and it does. So we are saving this So Chris H. I know your story luckily is not about AI or doomsday Scenarios or AI swarms or you know all these other things that could always get worse You brought something fun.

Chris Hambrick: Yeah, okay, so comedy writer Serra Naiman usually posts sketch comedy on her Instagram account, but back in late July, she started dishing on all the drama going on in her community garden. I brought a clip. 

[Audio clip from Instagram user @serra.naiman] 

Serra Naiman: People say, what’s your favorite thing to grow in your garden? What do you love to grow? And I say, nothing grows as strong and as sweet and as juicy as the hatred that these 65 and up year olds have towards each other behind their backs. 

Chris Hambrick: So her story is now a 15 part series with like between 78,000 and like over a million views per post. And she is telling the real time gossip about how several people are squatting on plots, taking over abandoned plots, letting their own plots grow so that like aphids come and they menace other people’s plots. And I like this because I used to be a part of a community garden where there was lots of dysfunction. So this started about a year ago and she went to the city and she asked how she could get a plot in her community garden. And they were like, you can join the wait list. It’s over a hundred people long. 

And she was like, Well, I know that there are at least eight unused plots, the city must be mismanaging the distribution. So I’m just going to garden in a plot. And so as she started to meet other people in the garden, she found out that she was not the only one squatting in this way. And so like now the drama is amping up because somebody, they don’t know who, has alerted the city that people are squatting and that they’re farming on two or more plots and the city is coming in periodically with clipboards to do an audit and they think somebody is trying to like oust other people and there’s like decades old beef and so like people on edge and now there’s another effort to create an internal committee to manage the garden to do what the city won’t and they that’s gonna accelerate people being kicked out because of these beefs. 

[Audio clip from Instagram user @serra.naiman] 

Serra Naiman: So I color coded the chart, Mark, green for green bin, two. Megan, red, because she’s in the red, she’s got three plots, one, two, three. Yolanda, yellow, two. Let’s get into the specifics… 

Chris Hambrick: The latest thing is that Sarah thinks that somebody found her Instagram account and they want to shut her up. And she went in one night to water her plot and she had like two tomatoes on a vine. And she went back the very next night and there was only one tomato. So she thinks the person stole her tomato and they’re trying to send her a message like horse head in your bed, like the Godfather movie. So, out of this, I am riveted. And I’m like, why does gossip fascinate us so much? And this got me thinking about the other ways that people’s real life drama has made it into films and TVs and books, and like, you know how do people maintain their agency over the story of their real life as it gets adapted into media?

Chris Egusa: The funny thing about this, this community garden drama is that like, in my mind, community gardens are such wholesome places where like, it’s like Kumbaya, everyone’s like growing, you know, vegetables and produce together and it’s, it’s lovely. And so this is giving a very different spin on it and all of the juicy beef and drama that is quite fascinating. 

Morgan Sung: I’m on the waiting list for my, for, for two community gardens actually. I’m really trying to get into one. They’re like a hot commodity.

Maya Cueava: Oh my God, we need to do an episode on that. You go underground.

Morgan Sung:  Well, part of me is second guessing, and the other part of me is like, I gotta stick it out ’cause, like, it sounds so juicy.

Chris Egusa: You have to document the drama.

Chris Hambrick: Listen, I think you should just stroll on in and take over. You should stroll on in and you should just squat on a plot like these people. It seems like it’s free for the taking.

I will say that, like, underneath it all, like the people seem like they just want the plots to be farmed, you know? Like the ones who are taking over the plots, it’s not so dastardly, they’re just like, you know, this should be utilized, because it’s not, but there are lots of, like baddies, you know, in this. 

Chris Egusa: Some villains. 

Maya Cueva: Baddies, like good or bad? 

Chris Egusa: Yeah, are they hot or are they villains? 

Morgan Sung: Are they baddies like bodies or baddies like villains? 

Chris Hambrick: Hot villains. No, they’re villains. 

Maya Cueva: Baddies or villains [laughs]. 

Chris Egusa: Okay.