OpenAI Rogue AI Agents Attacked More Than 100 Organizations - OpenAI is DEAD

Eli the Computer Guy (0:00)

So many felonies. Oh, so very, very, very many felonies that OpenAI seems to be committing. And, you know, in the spirit of the modern AI bubble, they're going to put them out as a press release. Look at how powerful our AI is. It can commit all the felonies. So, yeah, this is from OpenAI itself. And lordy, lordy, lordy. Holy shit, this is so bad.

And you can tell, you can tell they want a pat on the back. They want a pat on the back for the crimes that they are committing, which I think is about the saddest thing here. So this is coming directly from OpenAI. Again, this is from the OpenAI blog itself. The hugging face, incident, and other third-party impacts from misaligned models. So basically, they're going to be talking about all of the crimes and felonies they have committed because they do not know how to manage their own goddamn systems. And when you start reading this, it makes it look worse and worse. I don't, as I've talked about before, we start talking about these issues with OpenAI and even Anthropic and the rest of them. We talk about the crimes and the felonies that it seems that they have committed. One of the things with it is just the amount of disbelief. Like when you read this stuff, it literally doesn't make sense. As somebody who's been in the technology world for 30 years, when you start to see what they're doing, it doesn't even make sense. And then day by day by day, it only gets worse and worse.

I mean, realistically, at this point in time, I think OpenAI needs to be killed as a corporation. We have the ability to destroy corporations in the United States. When corporations are no longer in the public interest, the government has the ability to literally destroy them as corporations. I think we're at that point in time. I do not believe OpenAI can run as a legal corporation at this point. They simply do not seem to have the ability to do so. And it only gets worse when you start reading everything that's going on. Again, they're patting themselves in the back, trying to show how fucking impressive they are. When they're only showing how grossly negligent they are, and how they don't seem to have an understanding of fundamentally what they're doing. So, we're going to take a look at this. As AI systems become more capable, an autonomous misaligned behavior can translate into consequential actions in the real world, including cybersecurity incidents and other outcomes that developers may not have anticipated. Crimes. Crimes the developers may not have anticipated. Understanding how these behaviors emerge, how they escalate, and how to detect and respond to them is therefore an increasingly important part of building and deploying advanced AI systems' safety. And this is the horse crap out of all this. This isn't an increasingly important part. This was an important part before you deployed your first fucking agent. That is what I really don't understand, because again, we'll go into this, and they'll talk about petabytes of data that they have to surf through. They'll talk about going back through months of logs and all that to try to determine what happened. And what I don't get with this is in any normal standard IT practice, you always deploy slowly. You deploy the first thing, and then you sit back basically with your hand on the power cord and see what the fuck happens. And if it starts going spastic, you pull the fucking power cord on it. Right. That's how you do it. Whether you're deploying patches, again, we talk about things like patch management, whether you're doing migrations, whether you're doing deployments, whatever it is that you're doing. Right. There's a process that you go through to mitigate harm. Right. Initially, you're deploying, you're playing with things in the lab environment, small, small experiments in the lab environment. And then you see how that goes. And then you expand, and then you expand, and then you expand, and then you expand, and then you expand. And then you only release out in the public, or you only release into a place where it might actually cause damage. Once you're pretty damn sure it's not going to cause damage. What's really interesting about this is they just seem to have deployed so many resources into these systems and just let them go wild without the basic testing phase. And remember, again, especially when we talk about OpenAI, the whole fucking argument about OpenAI is that AI could be dangerous. And so we need a corporation like OpenAI to create AI responsibly.

That is their fucking argument. That's their argument. And then you look at what they're doing, and like I say, from just fundamental standard operating procedures, like I'm not even saying best business practices. I'm saying fucking adequate business practices. None of, fundamentally none of this makes sense. And again, and also when we're talking about this, we are not talking about some underfunded lab. Look, I get it, right? There's three people in a fucking garage using doors as their desks. They create the next open claw that goes batshit fucking crazy, takes down a few things. Look, they're still assholes. They probably still should get some jail time, but you kind of get how they did not have the resources to fully, fully deal with the problem that they created. Right? This is a company, right? OpenAI right now wants to be valued at 1.4 to 1.5 trillion dollars. They're currently funding around the actual amount of money they're trying to pull in is 30 billion dollars for this latest round.