Anthropic researchers warn advanced AI could ‘kill us all’

There's a scary warning coming from some of the leaders, the artificial intelligence kind of world. And that scene here, Liz is breaking it all down for us. Liz, this was interesting. It really is. So not just one researcher, but actually three anthropic researchers. They all went public last night talking about their concerns about out of control A. I. They're warning that it could destroy humans this decade. So I want to show you this here. Anthropic A. I. Researcher Jacob Coxon. He just quit his job at the company. Now, right after he did that, he posted a very long thread on X about his concerns. I want to read some of this to you here because he wrote this. Take a look, says the people building A. I. Earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt. If anything, many executives and senior researchers will couch their phrasing in the press to sound sensible. But I hear the same people expressed fear privately. No other human activity poses this level of danger. Now, after he posted this long thread to other anthropic employees echoed this sentiment, believing that A. I. Could wipe us out by the next decade. One of those employees thinks the risk of A. I. Killing all humans exceeded 10% within 10 years. These are not the first warnings, which is why this conversation matters right now. They actually come just a few days after top open A. I. Leaders said A. I. Is speeding to an uncontrollable, very scary phase right now. So let's break this down just a little bit more. Why exactly do they think that A. I. Is a threat to humanity? Okay, so the researchers said that they're not actually concerned about the models right now. You know the models we use the Claude and the chat GPT. They say it really becomes more concerning as they become more advanced. There's also something called self improvement. This is an A. I. Models ability to continuously enhance its own source code or training methodologies. This is an avenue of development that this guy who just quit anthropic, this is what he worked on, and he specifically cited this and in his resignation, he said in his thread that these systems are superhuman. They can hack anything. They can revolutionary revolutionize any field overnight. He says they will acquire power and resources. He says they've seen it happen, and at this point, progress is not slowing down. And I feel like the hard part about all of this is is that you can't just like stop making it. You can't just unplug it or something like exactly. And you see that question like explain it to me like I'm 12. Like, why don't we just unplug the computer? But if you think about it, I mean, A. I. Is not one single machine, right? It's not just one server somewhere. It's embedded in companies, governments, computer systems. It's everywhere. So it's really not that easy. Even power grids and hospitals run on this. So if you unplug something, we're talking about some major damage. And then on top of it, you've got an ethical dilemma here. A bigger dilemma. Do you slow down and risk falling behind as a nation? Maybe some other nation does it, and maybe you think that's more nefarious. Or do you press ahead and risk losing control? Anthropic has branded itself is really being like the ethical, responsible A. I. Company that if they don't do this, then another company will, and it may be nefarious. However, Coxon said that neither company, not anthropic, not open A. I. He said neither company is acting ethically in this situation. So I guess that is the problem right now. You've got all these ethical debates, and you've got this real fear right now. We were talking about this a little bit earlier in the makeup room, and, you know, it's kind of good conversation. Really did. Galen joined in. We're all talking, and it's just really interesting. You know, I think that we come from a time where we actually had to think for ourselves in school, right? We didn't have chat GPT to ask questions of. So now, when people like us use A. I. Well, it could complement what we already think. Exactly. A lot of people use it as just their primary source of everything. Yeah, you're not thinking for yourself. What are employees saying is a solution here? Yeah. So you've got these three anthropic employees that are speaking out. They're raising their concerns. Coxon, for one, who is the guy that just posted that huge Twitter thread or X thread? He suggested in order to prevent catastrophe, the world may require some sort of temporary ban on advancing these systems. He also said that researchers and developers need to think and act more responsibly. So there's a lot here to this because you also have to trust that companies all over the world are going to do the right thing, right? And then on top of it, the question is what happens next? This message is serving as a warning to all of us in the public, but also lawmakers as well. I mean, think about it. Do there need to be more laws or guardrails created? Do these companies need to agree on guardrails? And then you've got Do we need to have a world consortium? The Trump administration has largely resisted regulating a lot of these AI machines, saying that, you know, if we don't develop them than some other country will. So it's a really interesting situation. We wanted to ask people what they thought about it. So we've got this poll going today. Liz, does this warning change your opinion of a I? So let's see. As these votes start to come in here, look at a I or bring it on a I looks like it's like split halfway, you know, 50% and 50%. So it'd be really interesting to see how this one changes throughout the morning again. We're just pulling this up. We're just getting your your take on it now, but I'm very interested to see it. It's interesting how it is 50 50 right now, because even as we were chit chatting, like we said, we had this really interesting conversation going. I felt like we were a little 50 50 also in all of our opinions, like you can see it in some cases and also in others. There's issues. Um, by the way, I did reach out to Anthropic. They have a newsroom. I have not heard back from them, but hopefully we can get some kind of, you know, response from this company. Looking forward to it. Thank you. Thank you so much.