Anthropic AI Sets Up Biology Lab for Physical Experiments - Dario Is Building Bio Weapons

Eli (0:00)

Oh, mommy.

Mommy, I'm scared.

Oh, here's a thing. Here's a thing about the boogeyman. The boogeyman isn't actually real. Your mommy can come in, she can tuck you into bed, she can give you a kiss on the head, and she can make you feel better. Because realistically, even when you're a kid, you know the boogeyman isn't real, and so that level of comfort will allow you to get some sleep at night. See, the problem with Dario is that motherfucker's real. He's real. He's real, and he really is trying to kill us all, from what I can tell. So, see, this is both an interesting thing and a bit of scary thing at the same time. Why I find this to be scary, and I'm not joking about this one, is because I don't get Dario. Dario comes out, and he talks about AI being more dangerous than nuclear weapons. Oh, shut up about how dangerous this shit is. The employees for Anthropic literally come out, and they say there's a greater than 10% chance that AI will kill humanity. And I really want you to think about this for a second. They're not saying that there's a greater than 10% chance that AI will cause a mass casualty event. Right? Oh my golly, a thousand people died because of AI., right? That's horrible right there. No, they're literally coming out and saying, no, they think AI will literally wipe humanity off the board. Not destroy the United States, not destroy the Western world, not destroy white people or black people. No, humanity. And they still come to work every day and drop more models. Open AI is literally apparently going to drop a new model before their IPO after Dario has said that AI is so dangerous that they need to slow down development. What the fuck?

This is not an AI doomer in his basement screaming about how dangerous AI is. Because that would be stupid. It's not that dangerous, right? But they, the people creating the AI want everybody to know, it might kill humanity. And then we find with Anthropic that the Houthis, the Houthis have literally used Anthropic in order to build ballistic missiles. And one of the things that Anthropic comes back with is like, well, we stopped them during the process. Like, we found out after they launched at least one ballistic missile test, and they were trying to analyze the problems with that particular ballistic missile, that is when we terminated their account. And they're like, they're patting themselves on the back. So you're like, so you're patting yourselves on the back because you only help the Houthis create 75% of a ballistic missile instead of 100% of ballistic missile. So you're patting yourself on the back because you help the Houthis in the most difficult part of the engineering process. And now, once you block them, they just have to refine the errors that they found, right? Like literally, this is, this is anthropic patting themselves on the back. They also came out, and again, patting themselves on the back, talking about how people were trying to use anthropic in order to create biological weapons. Again, you can Google search this. Don't listen to me. Listen to Google, all right? And then, like literally, it was talking about how they found people trying to use Claude in order to create bio weapons, and they stopped them. But also do you remember, we've also talked about before, where anthropic doesn't really seem to have log files that are worth a damn? Again, I did a video about that before, right? Where people's Claude accounts are getting compromised. Basically, all they want is log files to be able to audit what happened, to try to understand who is using their accounts, or optisticially. And apparently, anthropic doesn't have those log files, literally. We talked about this before, where anthropic apparently hacked some particular company. Yeah, here's the thing. They hacked that company in February. They didn't find out. They hacked it until August. Not only do they not have real-time visibility into their systems, or near-time visibility into their systems, or, you know, fucking snail time visibility into the system. It took them eight months to find out that their systems had committed felonies against other companies. And this is just like literally how bad they are. And, you know, with all of this coming out, Dario is like, well, you know what we should really do right now is drop a new model, because that will make it better, right? Know your customer. And I do want to say this. Know your customer would solve a lot of these problems. Know your customer is something in the financial world. Basically, if you go to create a bank account in the United States right now, you have to come with government and ID, all of that kind of stuff, basically track the account back to a real human being. You know, if Dario didn't want to destroy... So Dario believes his technology has a real chance of destroying humanity. And instead of implementing real and actual processes that could prevent people from possibly creating bioweapons or whatever else, he doesn't want to do that because that might affect his $2 trillion valuation. Anyway, so we have all this. We have all this going on. And now we find out that Anthropic has quietly set up a biology lab so it can do its own drug program.