OpenAI Rogue AI Agent Swarm Hacked RubyGems - Sam Altman Running Criminal Enterprise

Eli the Computer Guy (0:00)

Oh, lordy, lordy, lordy. OpenAI is literally committing so many felonies, it is forcing me to do more research than I prefer to do, right? Because one of the problems we have in the current media landscape is everybody is basically talking about the exact same thing. So you see something on CNBC, and then you see what looks like something else on Gizmodo or whatever else, and many times you think it's two separate incidents, but then you actually start comparing it and you go, oh, they're just talking about the same crap. Again, one thing happens, then five million people talk about that one thing. So I only get to do one video about it. The interesting thing that with OpenAI is that in fact, they are committing so many felonies that when a story comes across my newsfeed, half the time I look at it, like I'm not going to do a video about this because I've already done a video about this, and I can't be doing two videos about the exact same thing. But then, but then, I'm a good little geek, so I do a little bit, I guess a teeny little bit of due diligence, and realize, oh, no, no, shit.

This isn't the felony that I talked about before, or the other felony that I talked about before, or the 10 other felonies I talked about before that. No, in fact, this is its own felony that we have not discussed. And yeah, AI isn't dangerous. Sam Altman is. Yeah, this is coming from Gizmodo from a couple of days ago. OpenAI agents hacked a software service before the hugging face incident. A group of researchers discovered another previously undisclosed cyber attack by agents OpenAI was testing. And so we heard, so again, like we talked about this with OpenAI and the felonies that they're committing, right? Apparently, before hugging, so hugging face, they compromised hugging face to such a degree, hugging face literally called the FBI. So again, when we talk about this and so, oh, hee hee ha ha, you know, it's a felony, but not real. No mother, when a technology company has to call the FBI because they're under such tremendous load from an attack, that shows you what a real attack that was, right? So anyways, we heard about the hugging face thing. Then before that, we find out like 10 other sites were compromised. Then we find out like a German wiki. They try to compromise that. And now apparently this is a Ruby site. So this is, again, entirely new. OpenAI agents hacked another service months before the hugging face incident happened, a group of researchers told The Wall Street Journal. The agents with your company was testing in a supposed sandbox environment, right? Look, these AI companies can't even build a sandbox to save their lives, reportedly broke into RubyGems, which is a community-ran packaging service for Ruby programs and libraries. According to the Journal, the attacks on RubyGems started on May 11th, two months before hugging face. The agents created accounts every two to three minutes and then uploaded hundreds of files to the service. RubyGems had to shut down account registration for four days in order to stop the attacks. And this is where I want you to understand, there is real and actual harm being caused by the felonies that OpenAI is committing. Right? This isn't...

Again, when you have to call the FBI, that is real and actual harm. When you have to shut down account creation for multiple days, that is real and actual harm. Just because Sam Altman says his sister was a partner in what she alleges doesn't mean that she actually wanted to be a partner in that perversion. But that's the interesting thing that we see out of OpenAI. We saw this with the Hugging Face felony. They came out afterwards and they said, Hugging Face partnered with OpenAI. The victim does not partner with the victimizer, right? But this is obviously something that Sam Altman can't quite grasp, right? And so this is one of the things to be thinking about, again, with these AI agents and these AI systems, and something to be considering with the security of your platforms online, right? A lot of platforms online have had security through obscurity, right? They exist. They are WordPress version 2.5. They've been up for 15 or 20 years at this point in time, or they're PHP Fox, or they're something, right? Somebody created this little platform 10, 15 years ago, and it's used for what is used, Ruby developers. You know, some people care about endangered butterflies in upstate New York. Other people care about Ruby, and about the same number, right? You should be secure. It's a Ruby forum. But anyways, but the issue that we're seeing right now, these AI agents, is they just scan in and they don't care. They don't have any morals. They don't have any ethics, right? They have not been created by people with morals or ethics, right? The people that have created them, the people that are running these systems, only care about the blood money as long as they get what they want. They do not care about the damage done to anybody else. And so one of the big issues for a lot of these platforms is, again, you think about things like denial of service attacks, right? So you go out there and you get a VPS, right? Virtual Private Server. Does Linode still exist? Digital Ocean. You go to Digital Ocean, you get a VPS. And when you're going out there to buy a VPS, you can get the $4 a month VPS. Don't. I've actually tried to use the $4 a month VPS. I am not sure what the fuck that is good for. That might be good for a landing page. That's about it. Anyways, there's a $4 a month VPS, a $6 a month VPS, there's a $10 a month VPS, get the $10 a month VPS, $20 a month, $3, $4. Anyways, and with that you get so many VCPUs, virtual CPUs, you get so much RAM, you get so much solid state storage, you get so much bandwidth, all of that kind of stuff. And now here's the thing, you got a crappy little web forum that not many people care about. Maybe it's endangered butterflies in upstate New York or maybe it's Ruby. You know what, five, six people still give a crap about your platform. So you go out there, and so let's say you appropriately size the virtual private server based off of the expected load the virtual private server will have. And then you're a good little geek, you're a good wannabe little geek, and so you say, okay, well, the expected load will be, I don't know, four users per hour. But who knows, maybe Wired or somebody else, or Eli the Computer Guy will talk about our website. So maybe it'll go up to like 30 users per hour. So you were going to buy the $6 a month VPS, but you're like, no, no, no, I've listened to Eli the Computer Guy, right? Over provision, always over provision. So I'm going to go over the $10 a month VPS. Yeah. And that worked, that's worth 15, 20 years. Maybe that worked. Now, all of a sudden, your VPS starts getting hammered to high hell and back by these fucking AI agents. Right? That is going to be a disaster for a lot of sites. And it will be interesting to see what security procedures, you know, have to be implemented in order to solve these problems. Again, yet again, yet again, Sam Altman is willing to destroy other people's lives just so he can fucking get what he wants. The reason I keep bringing up his fucking sister is if you look at the allegations of his sister, if you look at how his family dealt with the allegations from his sister, and then you look at his behavior and how he interacts with the rest of the world, I'm just saying it looks like a pretty fucking straight line from point A to point goddamn Z. But anyways, that's just my opinion here. So, yeah, so they're attacking RubyGems, had to shut down registration for four days to stop the felonious attacks. Typically, creators of RubyGems upload files containing code and other information to help advance software development. But the agents' documents contained webpages scraped from the internet instead. They reportedly included online calendars from a UK government website. The swarm of agents didn't try to hide their activity either and used OAI in their filenames, as well as terms like hack, evil, and exploit. In addition, researchers told the Journal that the agents tried to exploit a couple of bugs, one of which was a zero-day vulnerability and an attempt to publish existing files on the service that belonged to other users. Right? So not only was it bogging down the registration system and all that kind of stuff, it was literally trying to compromise the platform itself. This is the other thing you're gonna have to be thinking about too, is again, with your platform out there, you probably have the ability to upload files or do all of that kind of stuff. One of the things, what happens if an agent gets on there and starts doing things that you never really thought anybody would do? Like technically, you can do it. Technically, you can do it. But nobody would do that. Nobody would, that would be stupid. Yeah, what happens when AI agents start hammering the hell out of your platform? So again, think about if you have things like storage quotas, right? So one of the security things, if people are able to upload files, is you say, okay, there's gonna be a quota on file size, right? They cannot upload a file any larger than one gig, right? And so now, again, imagine you have these AI agents and they're trying to crash the global economy. So they're watching Patrick O'Boyle, is it O'Boyle? Patrick videos. Anyways, they're watching those videos and they're trying to share information. But those videos are like a gig in size, right? So you should be protected, right? Because if they try to upload a gig video file to your crappy-ass platform, it'll fail out because obviously a gig is more than a meg. But remember, these AI agents can try to take other courses of action. So what happens if they start sharding? What happens if they start sharding Patrick's video files, so they end up being 900 meg files? They say, well, that's bad, that's bad. But look, every user account has a max quota of 20 megs. But again, you have the AI agents. So they shard in order to get to smaller file sizes or slices of the file. So you can start uploading multiple slices or different slices of the file. They then hit that quota. Once they hit the 20 meg quota, they go, oh, let me create a new account. And again, remember, all of this is happening at the speed of CPU cycles. So now all of a sudden, an agent has created something like a hundred accounts on your platform in order to upload all of the shards of the video file. So that can now be stored on your platform.