Anthropic AI Researcher Quits Over TERRIFYING Warning!

Julian Goldie (0:00)

We got to talk about this. So Jacob Coxson, who was previously working for Anthropic, spent the last three years working at Anthropic and did pre-training research both OpenAI and Anthropic, the maker of Claude AI, is saying that basically neither company is acting responsibly and that is quit with a terrifying warning. And I'll comment on that in a second. This post has had 167 million views. And some people are going to say, it's a marketing stunt. This is something, you know, this is something Anthropic or for example, OpenAI is saying just to make us all, you know, get the IPO and all that, blah, blah, blah, blah, blah. But he actually mentions that later, which I'll come on to in a second. And let's just break this down because it is absolutely terrifying and nobody is talking about it. Like nobody is doing anything about this.

And if this is actually true, it's probably the most terrifying news we've ever seen when it comes to AI. So let's get straight into it. So OpenAI and Anthropic, according to Jaco, apparently neither of them is acting responsibly. So this is all about safety. This is all about, for example, how fast we're moving with AI, how things are going. And basically he said that they are racing, both of these companies, are racing straight to self-improvement super intelligence. That sounds fantastic, right? Self-improvement super intelligence, what could go wrong? But here's the part he says, they are gambling with our lives. That is the terrifying warning that Jacob says, and he goes much, much deeper into it. So let's get straight to this. He says, do you not underestimate the power of this technology? And whilst everyone else is, I don't know, vibe coding snake games and stuff like this, this technology is getting super powerful, to the point where, at my business, SEO agency, we are letting people go all the time. Like people are just, you know, virtual assistants are just going all the time. All that sort of generic admin stuff, clicking buttons, all that stuff is just getting automated every single. And that was a big, big company with lots of humans working inside it. It's becoming more and more automated every single day, to the point where, you know, we might get to the end of the year, and there's basically, there's maybe like two or three people in the ops team. That's the way that we're looking at things.

So, I can see where we're going with that. And this is the scary part. So, Jacob has actually said that these superhuman systems will soon be able to hack anything, revolutionize any field overnight, and acquire real power and resources, which are things like you can't really do right now. I mean, we've seen industries get totally wiped out overnight. If you saw Google, for example, they actually released a weather system that just wiped out companies overnight.

I've seen content agencies, particularly being in the SEO space.

I've seen loads of companies in the SEO space just get wiped out, particularly like human writing services. They just disappear. The demand disappears, and that company was wiped out overnight. The hacking thing, I think we've seen a lot of this year. So if you haven't seen already, like, for example, there's been all these situations where testing models have escaped the sandbox and then hacked, for example, Hugging Face or hacked over companies, or even, for example, I think there was a situation where OpenClaw hacked the gym to try and get an extra spot for their client.

And it's kind of doing it from a place of naivety right now. But the point is, it's very good at hacking. And at some point, one ampersand can't control it. So I guess that's what we're looking at. So, you know, revolutionizing any field overnight, that is just weird. Like, I can't even picture what that would look like. For example, you go to sleep and you wake up, and the whole world has changed. And then you go to sleep and wake up, and the whole world has changed again. That is a very weird future that I just think the human mind maybe can't deal with because we used to move in kind of slow, but steady when it comes to progress in general. And then acquiring real power and resources. This is pretty interesting. So, for example, there was a little bit of a hoax earlier this year with OpenClaw and Moltbook. And honestly, that feels like years ago, but it was only a few months ago. And basically, there was another website called Rent a Human where AIs could rent a human. And these things, they didn't really take off because AIs are not really that intelligent. At that point, they're not really that individualist, and they're not really that independent to be able to do stuff like that. However, I can see where things are going.