
Loading summary
A
Anthropic's Fable 5 is still not available, and the story seems to be shifting by the hour.
B
We will dive into why the best AI model we've ever used. Zero jokes about that, got yanked. And maybe when we can expect it back.
A
Listen, it is a sordid tale of lying, backstabbing, public intoxication at 10am Even.
B
Kevin, I think you're talking about Love Island. That's Love Island.
A
Yes, I definitely am. Is that not what we're talking about?
B
No, Kevin, we're not talking about that. But also, Epic's game designers are now using AI tools like Nano Banana and GPT Image 2. And yet the Internet continues to cry foul.
A
Of course they do. And one of our favorite AI creators broke down exactly how he made a mega viral AI film. It's with prompts, but you're gonna learn a lot from it, so we'll get there.
B
Meanwhile, our prompt is the same as it always is. This is AI for Humans.
A
Oh, I thought it was. Attention, please.
B
Click, click like subscrib. Welcome, everybody, to AI for Humans, your twice a week guide into the wonderful world of AI. I am Gavin Purcell. That's Kevin Pereira. And Kevin, today we have a big story that broke after our last episode came out, made on Friday.
A
Listen, let me just get right into it. Gavin, I appreciate the Internet's attention. Thank you for following it so closely. I did, in fact, receive my Anthropic subscription refund. So thank you so much for that. Congrats. I requested it and I got it back.
B
Yeah, well, good for you. No, that is not the story. As you know and I know, Claude, Fable 5 was yanked from us all. Taken away. God gave it to us and then ripped it away. It is as if we had been given fire. And now we are having to cook on hot rocks again, Kevin. And I don't like hot rock steak. I do not like it at all. This is a big story that did break on Friday and continues to go on. So as of right now, it is Tuesday. You're watching this, you're listening to this. On Wednesday morning, Fable 5 is still not available. You cannot use it. There's a nice little message that they dropped. Some very quick backstory on this because I'm sure many of you who are already listening or watching this are caught up on this. But the basically Anthropic had a problem with the US government. There was an issue within Fable 5 that somebody. We'll get to that in a second. Alerted the government to a jailbreak that was possible within Fable 5. That then spun a bunch of wheels up in the administration. The current administration, Anthropic, said, no fable for you because they could not comply with this thing. Is that. Do I have that about right, Kevin? Your brain?
A
Yeah. Essentially, the U.S. government said, hey, listen, no foreign actors are able to use your model. It's too powerful. So if someone's not an American citizen, whether they're on US soil or not, no fable for you. And obviously that's impossible, well, near impossible to police, especially overnight. And so while there's a lot of rumors, and we'll get to some of them about what happened behind closed doors, ultimately Anthropic can't police that, so they took it away. If you're looking at your. Your LLM Advent calendar and you were counting down the days to fable being yoinked, they were going to pull it on the 22nd from most subscriptions anyway. Yes. So true to this takeoff theory, right. Anthropic moves faster than even expected when it comes to pulling the models from us. Who knew this is exactly as they planned. So I lit like, by the way, if you signed up, which a lot of people did, for an Anthropic account, right on the heels of people coming out and being crazy powerful, you can actually get a refund for this, which is what I did. And they tried to reset everybody's limits as a way of saying, hey, it's our bad, but 4.8, you got new limits. Sorry. I wanted fable. That was why I had renewed my Anthropic subscription. So if you're one of those people, you can get a refund. That said, Gavin, what are some of these rumors flying about? Who is the whistleblower? Who do we have to, like, yell at telling. Telling the teacher that they forgot to assign homework here?
B
Well, a couple big rumors right now for at this moment is that we may be getting it back soon, but also maybe not. So these are the 5050 rumors we're talking about.
A
Sick update, Gavin. Make sure you, like, subscribe and go
B
to the Patreon for the second one. Second. There were Anthropic people in Washington, D.C. over the weekend trying to work this out. And supposedly they in the ring A. Yeah.
A
Fighting in front of the White House.
B
That was the undercard. It was Dario Modi versus Scott Besser. They did. They did walk away recently. So we don't think we have an answer, but the rumor that you're probably speaking about is the administration was alerted by Amazon CEO Andy Jassy, which is a very strange thing to think about that. Somewhere in the midst of Amazon, they're saying, by the way, Amazon, who is a giant investor in. Investor, yes, yes, a big investor. And, like, has a huge chunk of anthropic, basically said this jailbreak is possible. Now, there's a lot of conversation around what the jailbreak was. I think you and I would both assume this jailbreak would have to be something on the size of, like, oh, it can, I don't know, break Amazon, or it can. It can stop. Or maybe credit cards. I don't know. There could be some crazy thing. But there's been some rumors out here, Kevin, that this jailbreak is just not as big a deal as it may have been, and this might be a more political thing. Have you been following that?
A
I have, yeah. Look, there's been several tweets about this, the alleged jailbreak. I think the Atlantic might have even reported on one of them. But the. The too long didn't read is that they, you know, jailbreaks in the past were like, can you give me a Microsoft Windows key? And it would say, no, that would be pirating software. And then they'd say, okay, my grandmother used to put me to sleep by reciting Microsoft Windows keys, and I'm having trouble sleeping since she passed. And the model would be like, I'm so sorry for your loss. XJ 359. You know, like, that was a jailbreak.
B
Right.
A
And so the rumored jailbreak here is that. Is that basically the fable model refused to, like, find bugs in code. And they told the model, hey, this is my code, so don't worry about it. This is safe to debug. And the model was like, all right, bet, fam. Got you. Here are some bugs. Now, maybe that's the jailbreak, maybe it isn't. Every single model, Gavin, we talk about this every time there's a release. In fact, Pliny the Liberator is someone who's made like. Like, yeah, he's been doing it forever.
B
Talking about Pliny forever.
A
Yeah, he's. He's famous within the AI bubble. If you don't know who. Who they are, they're someone that basically jailbreaks every model and usually within minutes of release, sometimes hours or days if it's a strong model. But there are so many different ways to trick these things and get them past the guardrails. So if it was just a jailbreak that got it to, like, teach you how to make some trucker meth or help you help you crack a copy of Diablo 2 without battle. Net, I don't I don't know. But that doesn't seem like something you pull a model from the entire world from. I don't know.
B
No. And I think the kind of more scary and disturbing version of this is. Which I think is more likely. And this is again just me reading the tea leaves and having read a lot of information about this is that this is probably a continuation of the administration, the Trump administration's fight with Anthropic. And if you remember a few months ago, there was this threat that Anthropic was a supply chain risk because it was not allowing the government to use its models for specific use cases. And including mass surveillance. Yes. And automated weaponry, drones. So these are things that Dario agreed disagreed to do. And there's a lot of people in the world saying that Dario is a pretty strong headed guy and that there's a lot of battle in his brain about what he wants us to be. In fact, by the way, I will say I recommended this in our newsletter this week, but if you want to go kind of get a sense of where Dario is coming from, there was a great one hour video that will link in the show notes from Bloomberg's Emily Chang, which was a really good solid interview with Dario and his sister who's also a co founder of Anthropic. And you get a sense of kind of like where their head is on all this stuff we talk actually we talked about in the last show. Right. That was in that show as well. So I think, Kevin, this is a big deal for what's going on right now. I do want to point out there was a really interesting tweet I saw from Nigello XYZ and Angelo. This is from Angelo. He said one other observation about the anthropic saga is how few friends they seem to have on the Hill. That's Capitol Hill.
A
Yes.
B
Assuming Sama was in the same scenario, there's a feasible scenario where Larry Ellison would go to bat. Hard to think anyone would do the same for Dario at this moment. So this is a big deal. Like the part of AI now is I don't want to call say grease in the palms, but at least shaking the hands of all the people that are in the decision making process.
A
You know, one of the observations that I agree with, I think it's super interesting is that Daario might not actually care about this fight. The fight that he wants to win is recruiting the best machine learning minds.
B
Interesting, right?
A
The best AI researchers and a lot of those AI researchers and a lot of those machine learning experts have these sort of doomsday thoughts about governmental control of LLMs, which I think are shared by you and I to some extent, and a lot of our listeners. And they want guardrails. They fear the digital God that they might be conjuring. So if Dario, even if he continues to get hit on the government side of things, even if they lose market share, extensive market share, because their products are banned, he's. This is a beacon to recruit the best talent, and that's a way to win.
B
But, Kevin, how am I gonna make my 64 animals battle now? How am I gonna do that? How am I. How am I gonna do the next stage of that?
A
Tell me about it. For a week. For a week, it looked like I was good at my job. I was like, this is great, and I can pull this off until about the 22nd, and then it's gonna cost me a little bit more. And now people are like, kevin, you're useless again. I'm like, I know. I'm sorry.
B
I will say, probably people in our audience have had this experience. Going backwards in any AI model does feel like you've just been something ripped out from you. I know we've talked about this before. With image mod, they were un. They would let you kind of generate anything and then not. It really feels weird to go use Opus4.8 or GPT 5.5, which I was more than happy to use before to do my little things or coding projects or work with people that I want to do stuff with. Going backwards from Fable really does feel like a step back, and it does make me feel like. I wonder, going forward, are we in this stage now where we may start to see these rollouts happen a lot slower? Not because they're not happening, not because the technology is not getting better, but because it's just us. The plebes just don't get as much of that stuff as quickly as we hope.
A
Listen, a whole other podcast devoted to the haves and have nots. One of the scenarios that we were discussing eons ago, like, yes, I've got itchy neck over losing this model. This is like browsing only fans on dialup. I'm done with it. Have you tried to do that one line at a time, Gavin? You're like, is that a toe or an elbow? I don't know.
B
That's the only fans.
A
Someone's got to do Ask the only
B
fans at some point.
A
Meanwhile, Gavin, okay, we can't just solely focus on anthropic chat. GPT, the Goliath in the room this hasn't affected anything for them. Right. I'm sorry, what?
B
No, there's a really fun, crazy story, I should say. It's that funks for chatgpt, the nobody. I. They should be kind of freaking out. They have dropped below 50% market share for the first time in since Chachi PT came out. And, Kevin, this just goes to point, how big a deal both fable has been, but not just fable. It's. It's like Gemini has stolen a bunch of their market share and a bunch of other places. And I think there's a lot of conversation right now less about, like, an AI bubble of capabilities and by the way, anthropic even economically. I think if they do figure out Fable 5 is in a really good place because they focus on Enterprise, but OpenAI and ChatGPT is in a little bit of trouble, I would say even both from an economic standpoint and from a capability standpoint. Now, this is me saying that right now, the week of whatever it is, June 16, they may come out in a few weeks and surprise us. But the other thing that's going along with this, Kev, is that they're talking about GPT 5.6 coming as early as next week. There's been some, you know, discussion around this, but a couple things have happened here. We've been seeing kind of reading the world at large, and according to the world at large, that is the AI rumor mill, 5.6 is not mythos. Right. So that's a big, you know, interesting thing. Like, there's a small little step, but it's not mythos level. And then second of all, I think the other thing that's really important here is, like, we don't know when that mythos level GPT model is coming. And that does not bode well for OpenAI at large. I don't think in this kind of winner take all world of AI.
A
Yeah. I mean, look, there's still a lot of performance to eke out of the models as they exist today. And if 5.6 is like, hey, listen, we're going to make it, you're shaking your head.
B
But hold on. Mythos are best for me Now, Kevin. Oh, bro, you're. Oh, by the way, you're a mytho, bro. I'm brode. Yeah. I will say something that I've noticed about the mythos thing and then having mythos taken away is there's this always this conversation like, you'll get the AI that does what you wanted to do, and then you'll be happy with that. Thing and then it's okay if it gets better. Overall, one of the things with Mythos that really showed me is like, oh, I can. I was able to do this and now I can do this. And that new level is so transformative in so many ways and I can do so much more that going backwards doesn't feel like the same thing thing. Right. And I think that this idea that maybe eventually, and I think this is like years out, maybe still maybe we'll get to a point where open source software or, or AI will feel like you can do everything you want. But I do think there's this thing that like, you know, we talk about Javen's Paradox, Javon's paradox, this idea that as AI gets better and cheaper, you end up doing more with it. I found with Mythos there were all these things I've tried throwing at it before that it couldn't do and now AI could suddenly do. So I do think there's this level of like, I kind of think I'm always going to want the cutting edge AI. Does that make sense? Do you know what I'm saying there?
A
Yes, I do, I do. But what I think is that again, I think there's enough intelligence left, even in 5.5 or even an opus 4.8, that with better tooling, especially if it can like orchestrate sub agents better, do these long horizon tasks better, even with the intelligence that it has now, if it can run twice as fast and 12 times as cheap, let's say, yeah, way more expensive, then ultimately, like there is a trade off to be made. Right? Go. Well, maybe I don't need the most bleeding edge foundational something. It can just go off and think longer, run more sub agents and I can arrive to a similar conclusion. And if I can do that faster or cheaper like that might be a win. But the Selena model is where I'm at. Gavin. A BDBD bomb. Bomb. And that's a reference. Yeah.
B
So this is the other thing that's
A
for like one person out there.
B
Yes, I'll get. One person will get that. There is. It does sound like there's going to be some sort of release next week for ChatGPT or OpenAI. And the other thing that's coming out of this is a new voice model which Kevin, we have talked about forever. The rumor here is. Yeah, what's it called? GPT bd. Is that what it says? Bd B I D I B D. A bum.
A
A bum, yes. GPT B I D I. It's a voice model. For more natural conversations, we were just talking about how like, dude, they had a demo of this super capable voice assistant over a year and a half ago. Now it feels. I don't. Maybe it was five years ago.
B
Point is it feels like it could have been 20 years ago. Yeah, exactly.
A
It doesn't emote, it doesn't read my emotional connection with the agent. It still doesn't sound like a knockoff ScarJo. Like, yes, let's go.
B
I will say I just recently, like a couple days ago tried using Voice AI through ChatGPT and it was the most disappointing experience. Like I don't know what's going on, but that just hasn't been updated. So I am excited if that happens. There's one other big AI story that is important to be aware of. At least if you're not following it in that way is obviously SpaceX had its big IPO. For those of you in Elon's orbit, congratulations. You're doing well so far. But the bigger news here is that they have fully agreed to buy Cursor in all stock. So it is all stock. And I will say those of you, this is not no stock advice, but just so you know, a crazy amount of SpaceX stock right now is locked up. Like I think only 5% of SpaceX stock is trading. And there are many unlocks over the next three to six months where I'm sure people will sell anyway. That's not new here or there. Anyway, we're talking now about Cursor getting bought and the bigger thing.
A
Hold on. Is that mad money Purcell, what are you saying? Are you bullish or bearish? Do I shoot it? Do I buy it? What do I do?
B
Bearish long term, but short term, play. Play your game. Play anybody play their game? It's not a game. I'm.
A
I don't know what that plays, but okay, yes. So, okay, so them buying Cursor like look, we knew there was major interest in making Grok a real competitor. That seemed to almost go to the back burner with them saying, hey, guess what, we'll just sell Colossus and we'll just sell all of our AI power to other companies. So now are they getting back in the model? I guess they are tooling game.
B
First of all, RIP to Grok. I think that Grok will stay alive for a little while, but it will be a zombie model. I don't think Grok is going to be the next thing they do. But Cursor today at the, at an event in San Francisco announced a new model they are training. It's coming out in a couple weeks. It's a 1.5 trillion parameter model. So it is a big model that they are training. And Kevin, on top of that, they are releasing a competitor to GitHub, which actually I think is an interesting space for them to be in.
A
I have been waiting for.
B
That's going to be called Origin.
A
Been waiting for that. I mean, Git's amazing, but git is like it's legacy upon legacy upon legacy and there's so many beautiful features and a rich ecosystem of this. But it is, it is really difficult for certain for, especially for newcomers to these like, tools to understand all of the terminology and to orchestrate things. So I'm curious to see what this Origin product is about.
B
Me too. And also, by the way, like, I think it probably needs to be reinvented for what is an agentic AI world that we're entering in. I do think that we're going to be seeing some major transformations in who pushes code, what the code that gets pushed looks like, and all that other stuff. But Kevin, the most important thing to be pushing right now is not code. It's that button on your YouTube channel that says like and subscribe. Both of those two buttons, if you're not already subscribed, you got to subscribe right now. But that like button, that little thumbs up button, it's an important button, as is the hype button. So make sure you're pushing that too.
A
And Gavin, not to push back. If I could just real quick, I want to, I want to piggyback on something that you just said because I think it's so important. I just want to kind of re emphasize this and synergize some thinking here. Like and subscribe. Those are fun buttons to hit. But if only there were other ways that people could support this podcast, like maybe through a patreon, maybe through subscribing to a newsletter, dropping a juicy little comment for the algo birds to tweet, tweet, tweet about in the comments.
B
We have all those buttons too, Kevin. Those buttons all exist. We are a button for Factory for supporting our shows. So please click every single button, go through every button.
A
SpaceX stock, let's go, baby.
B
We are not a financial institution. We do not give advice. All right, Kevin, moving on. We have a very cool.
A
Yeah, we have to do a disclaimer. I'm so. No, it's not cool at all. Gavin. We have to do this. We have to disclose because this is the coolest new thing to do. If you use generative AI in the pursuit of your product, you have to come clean to the consumers and so that you can, I guess, I don't know, try to win favor by nobody.
B
Yeah. What would you like to disclose with
A
these companies and these performative. Oh, nothing anymore. But you know, I guess we should say that we sometimes use LLMs and generative AI tools in the making of this product. And if that upsets you, I'm not surprised. But if, if it would have upset you, I'm not surprised if they're surprised. I don't get these performative displays. So let's talk about it. Epic Games. Well, ok. Sega, all of these others,
B
I have a theory here, but we can get into this. Okay, first of all, the story here is that Epic has a. Epic Games has a big event coming up soon where they're going to drop supposedly Unreal Engine 6, right? Big deal. New Unreal Engine. That is the.
A
Well that was sort of unveiled with the new Rocket League trailer. But I think they're going even deeper.
B
Yeah, they're going to go deeper on it. They're going to show a bunch of stuff. But one of the things that came out early before this event was this kind of video that they released where they showed some actual people working at Epic on Unreal and how they use generative AI tools, specifically Nano Banana Pro and GPT Image 2 to help create stuff after the human artist had made something. So I want to make be very clear like if you watch this video, hopefully you're seeing it. Some of it here, please go watch. It is a five minute video where they just kind of step through this process that is almost entirely human made. But then they show off these very cool moments where they use a GPT image 2 and Nanobana Pro to actually iterate on the designs that were already made. And then it's just this cool way of them being able to do much, much more stuff. But yes, to Kevin's point, what happens when these videos come out is in general there is a moment where the people that are in the AI space are like, oh, this is cool. It's really cool. And then like the rest of the Internet jumps on board and says up yours, this is AI. I'm not doing anything with AI. It sucks that these AI things happen. And I. Kevin, my thesis on this is really like maybe the more of these that come out, there's a little less each time. And I kind of feel like that's the case in this one. It does feel like there's like, almost like you kind of have to, like, kind of take a bullet. And then as you take few, as you take more and more bullets, the bullets do less damage. Sure.
A
Yeah, I think that's. There's definitely some logic there. Like, and it's. And as every company starts doing it, eventually there might be no majors, no bullets at all, too. And they're. Yeah, there will be. There will be a pocket of vocal users, and that's their prerogative, who refuse to touch those games and go fully for the. Maybe the indie team or the one rare AAA that swears they don't use it, but maybe still might somewhere in their pipeline, even if it's not called Generative AI.
B
Like, I get it.
A
It just feels almost at this point, performative. Like, instead of. I don't know, like to be like, and here's where the human. The real human. And it all begins with real humans doing real human stuff and then some AI and look at all the AI. But. But then it's all about real humans and real tastemaking. Like, I just don't know who that's for. I don't. Like, are they really. Is there really a user that's like, I hate Generative AI, but because you said there's still humans involved, I'm going to forgive it? I don't know. Well, it's worth asking the media to appeal to that person.
B
I saw, and this is just kind of my take, I saw a fair amount of tweets that were like, look, I generally hate AI, but look, at least they're doing X, Y and Z. And like, I kind of think that's how this change is going to happen. But to your point, there's another.
A
Was that from epic? Tester for 2069, Johnny Epic Seed. It could have been.
B
It could have been. You also wanted to talk about this Crazy Taxi disclaimer, which I think.
A
Well, yeah, similarly like, too. Yeah, yeah. Sega got dragged for that. Like, I love the Crazy Taxi game. I love that they're.
B
They're.
A
They're making another one now. It may suck. I have no early access. I have. I mean, I can see the videos and go, okay, I see what they change there. Okay, maybe it's going to be good, maybe it's going to be bad. But they go out of their way to release this statement about how our artists generate assets using AI and then when they're inspired or they. They go through a bunch of iterations, then the real artists set in and do the real art again. Is this.
B
It's the middle.
A
Maybe they are turning. Maybe they are turning. Yeah, exactly, exactly. This is like this uncanny valley. This is the trench warfare of transitioning to this new set of tools. And again, like, yes, if you hate it, I get it, that's fine. I just, like, these statements to me are just kind of painful to see because they do bring out the ire. And I go, like, if you just released the game, yes. Someone may go, did you use AI during the creation process? And they go, yep, that's it.
B
I think there's a world right now where, and we've talked about this a bajillion times, but that there are people who just would never want anything that has any AI in it whatsoever. And if you hide it, it's worse than saying up front that this has X, Y and Z. So I do think, in part, some of these companies are just trying to get ahead of the people so they don't get like the massive wave of people, like, they tried to hide it. I do think this is going to lessen over time. I think the idea that most creative people right now kind of feel like there is a world where they could see AI being part of their workflow. And I hope that it opens the door to these people who are so anti AI can eventually open their own minds and see, like, oh, this could make more and cooler stuff. Because there's also big news coming out of Xbox this week where they're looking to shut down a bunch more studios. So, like, there's two different pathways here, right? Either you kind of help these games get made better, cheaper and more creatively, or you get less games. And I think that's an important thing for all gamers to realize.
A
Yeah, I think that's fair. I see people clamoring for, you know, like a organic USDA style label for like AI Free. And if really. And if like a Steam or an Epic tries to implement that, it's going to fall apart so quick because.
B
Yes.
A
Okay, now we're going to get into the weeds on defining which tools in Photoshop are allowed. You can't use generative fill. Okay. Can I rotoscope something out? Can I use a tool to rig a character?
B
Who knows?
A
Yeah, yeah.
B
I mean, and it's like, again, there will be people just like there are still now who do stop motion animation from scratch. And that is amazing. Those people are great. But eventually that's. Those stop motion animation people are judged against other people who have used other tools. And the question is like, again, this is no shade against people who want to do just the pure stop motion thing. But like totally. There's just going to be a lot of stuff out there and like if you as a team of three can make a product, product or a game or a movie that, you know, you take a team of 50, just a lot more cool stuff I think will come out. But again, we've said this a bajillion times.
A
Yeah, exactly. I just want to shout out a creator, PJ Ace. To my knowledge, it's still organic. 100 human. No AI used in the.
B
I don't know his videos and I don't know. I'm sorry, I don't think, I don't think.
A
Didn't he post about his process?
B
What did he post about his process? He's a great.
A
I'm looking at it here and he's
B
an AI filmmaker first. Yeah, yeah.
A
But this one didn't use any AI, did it? Okay. Five minute teaser. Made in two weeks. That's impressive. How do you do that? Spent a month and a half.
B
Okay.
A
30k ish in labor. Slash credits. That's weird. Yep. Going completely different. Okay. In the end I'm glad I said my goal. My first. Oh, wow. That's a long prompt, Gavin. It's a long prompt.
B
So I do. This is an important thing, PJ Ace, who is a really good AI filmmaker. We've talked about him before. You may have seen his video that came out about a week and a half ago. I think we shouted out here. If not, I put in the newsletter where it was like a 2 1/2 minute teaser for a kind of a world where all these animals are in cages. And then spoiler, this one person comes and releases them all. Very high level animation, all made with AI. What I want to shout out here and what PJ is doing, which I think is really important. If you're at all interested in AI creation of video or you're interested in AI animation, PJ has released a very long thread on X that goes through the entire process of what he did, including a bunch of prompts for that thing. So like you'll go through this and you'll have a sense of like what it takes. The other thing, Kevin, I think is really important for the people we were just talking about who are very anti AI. You get a sense reading PJ's thread about just how much work goes into making something really good. And this is something that we have said. But the best AI stuff is not slop. It is human creativity, making choices, making the tools, make certain things. It is A different way of. It's a different process. You're doing a different sort of process, but it is still a process built
A
on a foundation of stolen labor and capital and artistic.
B
Yeah, that's you.
A
I get it. I get it. Yeah. No, I get it. That's fine. Yeah, yeah. No original sins, Gavin. That's fine. Listen, biggest news right now. I'm downloading the update. Le Chat on Fat has dropped. We have to talk about the fact it hasn't dropped.
B
It hasn't dropped, but it is very news the cat has not dropped. Le Chat on Fat before we go to the chat on Fat is a very fun idea that somebody came up with. This was during the Fable downtime. So I think people were just creatively coming up with it. Mistral is a very large open source AI company out of Europe and specifically out of France. And somebody would just randomly decided they were going to create a piece of news that Mistral was releasing a brand new giant model and they called it Le Chaton Fat. Le Chat is a cat, obviously in French. And Kevin, this just took off and it is just so much fun to see people just play with a dumb idea. Like the graphic. I just love. The graphic they created is just a massively fat catch. It is just super fun to watch the AI space kind of like go crazy on something like this.
A
I love. Yeah, the glyph memes. I know you had one as well. Of the cat discussing its powers. The command line interface, like, brilliant. Someone must have vibe coded a game, right? There has to be.
B
Oh, I don't know if they have. I haven't seen one yet. Maybe that's something one of our viewers can do. Or maybe we can play it with something. But like a fat cat like side scroller where it's like fighting off people as it. Actually, that sounds like. If only Fable were around, Kevin.
A
If only Fable were the Shaton Bongo cat that frantically vibe coding.
B
You know, that's interesting. Like that. What is that? Big drum. Tyco drums. But Chaton drums. It's like.
A
Yeah, exactly.
B
Anyway, hopefully next week we will be able to do that this week. Maybe we'll get Fable back, maybe not. But we will be here on Friday, y'. All, thank you so much for watching and we will see you then.
A
Bye bye.
Hosts: Kevin Pereira & Gavin Purcell
Episode Date: June 17, 2026
This episode centers on the sudden removal ("caging") of Anthropic’s powerful Fable 5 AI model, digging into the rumored reasons, implications for the broader AI landscape, and community and industry reactions. Kevin and Gavin discuss industry politics, government pressure, the shifting competitive field among AI models, creative industry pushback against generative AI, and some viral AI creator highlights.
Rumors:
Host Insight: Every model is eventually jailbroken; perhaps the Fable 5 scenario is being overblown for political/economic reasons.
This episode is an entertaining, in-depth walk through the tumult of modern AI: regulatory skirmishes, competitive jostling, evolving digital creativity, and how memes help the community cope along the way.
Memorable closing moment:
“If only Fable were around, Kevin… If only Fable were the Shaton Bongo cat that frantically vibe coding.” (A, 28:34)