AI could wipe out humanity in a decade, researcher warns

Good morning, everyone. I'm fair warning on artificial intelligence coming from inside the industry, a former anthropic researcher voicing his concerns on what a I companies are doing with this. Very fast growing technology. Very fast. Yes So how about this? A I could kill all humans. That's excuse me. Okay, exactly. I was like, I'm like, Okay, let me sit down. Right this entire thread. Yeah That's the warning from current and former employees of anthropic. So Jacob Cox and quit his job at anthropic, he says, and that's an artificial intelligence company, one of the artificial intelligence companies, one of the top two. He then quickly voiced his concerns on what a I companies are doing with the fast growing technology, he says. Artificial intelligence could end the human race. He actually said there was going to be an issue, you know, before the end of the decade, and then I realized we're almost at the end of the decade. Anthropic safety executive Evan Hubbinger doubled down, saying, quote, We really do earnestly believe a I could kill all humans, saying there is less than a 10% chance in the next decade. People essentially are aware that the work they are currently doing if we don't slow down, could lead to a I take over by the end of the decade could happen as soon as next year. Some people even say six months. The whistleblower says part of the problem is that every country, including the United States and China, they all want to be the first in the a I arms race. So he's saying essentially, even if everyone has concerns, they're worried that another company won't have concerns, and they just want to beat them to it. Republicans and Democrats don't agree on much these days, right? But in this case, both parties do want attention on this. Democratic Senator Chris Van Hollen says the U. S. Needs urgent dialogue with China and Republican Congresswoman Anna Paulina Luna called for a special session on a I heard a researcher talking about this, and he equated it to it's called the paperclip theory, where if you give a I task, it will do the task to no matter what end. So, for example, if you give them a task of make as many paperclips as possible, then it will do that. And eventually, when it runs out of material to make paperclips, it'll start crashing planes from the sky. It'll start running cars into walls to get the metal to make more paperclips, not knowing when to stop, right, taking right, not knowing when to stop. Literally, that's right. Wow. So Governor Newsom agrees. Yeah. So he actually signed two major bills yesterday surrounding a I safety regulations. He told Politico there must be meaningful safeguards to protect the public, noting the concerns raised in recent incidents. And then what about anthropic? Are they saying anything about all of this? Right? Because you're like, Hello, talking for you. Yeah. Are you gonna say something? There is a bit of a response. They have been in support for months, actually, of what we knew Governor Newsom was going to sign. These bills were prepared for them in anthropic and publicly shown support for them, saying, in part, just yesterday, we continue to build models with some of the strongest safeguards in the industry, going on to say this work is also why we believe the world would benefit from the industry adopting a lawful, verifiable way to work together to pace how we release powerful models. Of course, you know, jokingly, though, it's almost easy to think we'll just unplug it. But these people who work on the product are saying it's not so simple. It's not so simple. Yeah. Yeah, it's definitely changing the world, and it's happening very quickly, which is, you know, very scary. Keep up with that. There are more immediate concerns. You know, when you talk about job loss and just even before getting to this, the world. Yeah, right. Yeah. Essentially, there are many concerns. Daily daily life. People worried about, you know, their faces being put on inappropriate videos and photos and how that can affect their lives. There's so many different concerns. There was a recent hacking thing where a bunch of I think hundreds of AI agents all worked together to solve this math problem. And it found out that they were communicating with each other using human language and almost going rogue where they were. Yeah, problem. Yes. Yeah. Right. Yes. So now they have their own. They found a way to go beyond what was the task. And yeah, right. And it was a scary situation. I think also even scarier. You know, you think like we're smarter. We're starting when you think about like a computer controlling a human. But like, raise your hands in the room here. If somebody that you consider intelligent has sent you an AI video or photo that they thought was real, right? Absolutely. Absolutely. Yes. Happens all the time. I fall for it all the time. I've been the one to send the AI. Yeah, yeah, yeah, exactly. And somebody I think is smart. That's right.