Anthropic AI Researcher Quits Over the Race to Super Intelligence

0:00a man named Jacob Coxon who is, doesn't have any followers on Twitter, no one knows who he is, starts an account. He's never tweeted before. Ever. He posts a tweet that becomes the most viral tweet, 165 million views, 776,000 likes, 200,000 tweets, and it sends shockwaves through everybody. And here's what he said. I resigned from Anthropic today. I spent the last three years doing pre-training research at both OpenAI and Anthropic. Neither companies acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below. Do not underestimate the power of this technology. These will be soon superhuman systems that can hack anything, revolutionize any field overnight and require real power and resources. We have all witnessed the progress in each of these domains and progress is not slowing the people building AI earnestly believe that it could kill us by the end of the decade. This is not a marketing stunt. If

1:04anything, many executives and senior researchers will couch their phrasing and depressed to sound sensible, but I hear the same people express fear privately. No human activity possess this level of danger. A common response. If they truly believe this, why are they still building AI at OpenAI? Many have not deeply internalized the civilizational stakes. At Anthropic, the stakes are well understood, but they are locked in a race to get their first. They believe no one else will act responsibly, so they must do themselves despite the risk. Accepting this race and entering the end game is hubris the gamble that should not be launched from a private company slack attempting to speedrun alignment should require extraordinary confidence that there are no better trajectories available. I'm optimistic about the potential for coordination. Warning shots like the hug and face attack have made pacing agreements between us labs more viable. I don't feel like we're on track to prevent a global race, which may require costly actions such as a temporary ban on improving model capabilities.

2:08If a lab researcher, if you are lab researcher, I urge you to consider what the next few years will actually feel like. Do you want to kick off super intelligent RL run without a rigorous understanding of its mind? Should you put your head down because it's happening anyway or take this moment to call for different conditions? Immediate reaction from everybody was like, oh my god, I cannot believe this is happening. You should read this. This is interesting. All of us, eras, this is something we should look into. Bill Ackman retweets. Very interesting. Very concerning. All this other stuff. And then all of a sudden, a couple other people come in and if you can pull up the tweet from Parker Thayer. Parker Thayer was the first one that kind of questioned this a little bit and said, hey, this post looks like the start of a very sophisticated and well-funded PR operation to get support for Democrats to regulate AI and to oblivion. Let me show you how it works. This guy with minimal followers, by the way, I think today he's got 260,000 followers. That guy named Jacob Coxon. Can you go to his account to see how many he's got? And then just come back? What's the number of? 260, 4,000 followers from 048 hours. If you want to go back to that tweet that we were on, this guy

3:13with minimal followers and no previous account activity goes to Wall Street Journal and publishes and exclusive with quotes from him on his resignation. 18 minutes before this post goes, a planning was clearly done in advance within hours. If the tens of thousands of reports and accounts 100,000 followers, the post is punchy, quotable and it seems professionally written. The first three accounts to quote, tweet within 15 minutes of the initial posting. Remember, this account had barely zero engagement. According to Groc, those accounts, Nathan, Peter, Nathan, General Counsel of Encoda, Peter head of policy at AI policy network and DK, whatever the next one is, is head of AI future projects, all of which are up and coming AI tumor policy advocacy nonprofits. The AI future project website says it's funded primarily by this survival and flourishing fund, which says on its own website that it has advised John Tullin Skype created one of the following lead investors and anthropic to grant $2.5 million to AI future projects. Anyways, if you go through it, you will see a pattern. Elan responds to this.

4:16It says Jacob Coxen doesn't have much of a resume, but we know that in 2022, he got a $20,000 scholarship from Longton Future Scholarship Program from Good Venture Foundation, one of philanthropic vehicles of Dustin Moskovitz, a notorious AI tumor who has spent tens, if not hundreds of millions on policy advocacy to regulate it. And so as you go through this, there's some ties to Sam Mangman, free, Elan responds back in Elan says, this is a bunch of BS. He marks the ex- anthropic staffer who warned AI could kill all of us in the tweet from Elan Mosk within seconds. Jacob Coxen responds back, as if Tom said he was ready for it. Brett Beard did a good job asking him a question yesterday. He said, hey, was this an intentional thing? Was there a whistleblower? Was there anything going on here? Here's what he had to say. Go for it.

5:26Yeah, so unfortunately, we know how to control nuclear weapons pretty much. We don't yet know how to control AI. This is possibly the most dangerous technology that humanity has ever created. It's basically out of science fiction. I think we have no other choice but to cooperate internationally, because an arms race would be disastrous in a way that no other human activity has been in the past. He asked him if you work with a whistleblower, he said, no, I'm not working with anybody. I'm not that this was not an orchestrator thing. So Tom, what were you at with this? Because we've seen this a lot right now with AI tumors. What do you stand for? What do you have to say here? After looking over everything and pouring over it and seeing the responses from people, this is a pile of crap. They have created the Greta for AI. This is the Greta Thornbrough for AI and that created it. Look at where the guy was. Parker Terry even points out. The guy was previously at news speak 2021. This guy has built up for this moment and he went to work for Anthropic for I think it's been confirmed now. Multiple reports six weeks and two days.

6:28I think in terms of actual calendar time, there's many reports on that. So I'm going with the six weeks. Really, why were you there for six weeks? And why only six weeks? Or were you there long enough to get your your beak wet on the data and then you could come out and do this? I think this is contrived. I think this is set up. I think there may be other angles in here that are trying to stop the Anthropic IPO. I think it has to pause. There are people over there that you and I know Pat that speculated today on X that they may have to delay the IPO and kind of let let the smoke clear on this while they're taking smoke. But that's what I think. I think this is contrived from the beginning. I think his history proves it and I think that that everyone from Elon must to others. You would expect to Elon to respond this way. But I think Elon is exactly right. And I think what you have here is the dooms and coincidentally coincidentally Bernie Sanders who wants to

7:28control something this week. And this week it is AI. I'm really I'm really well, I've said what I'm going to show this. Elon responds back to us seems like a setup and then Jacob responds immediately and says I'm real. And these are my real beliefs. You could ask your ex AI researchers about me if you hadn't fired them. So that's what he's saying. So to see where you at with this. So there are there's a political part and there's an intellectual part. The politically speaking ND UK we call people like Jacob Avocado eating soy boys. These are the younger generation where on the left where they have a problem with authority. They have a problem with work. They don't want to go to work. They don't want to be controlled by people. They want to be creative yet. We see time and time again as I said the likes of Greta again. They get controlled by the left ring established and you know they whether I directly or indirectly. So that's the political part and you go obviously covering it. There is intellectual laziness here that there's always been in every generation the doom and gloomy I personally believe that the future looks brighter. The reason

8:29for that is because today if somebody comes to you and says I've invented something it's a metal box you sit in it and it goes fast and you go to your destination and we're going to call the car. There will be some people who freak out. They would say this is the worst idea. It's dangerous but cavemen decided to explore the sea and at the time there were people for example saying well no we figure this is the end of the world because there's fish that are things that will bite us we'll drown humanity always finds a way to survive. We will be fine also if you really care you're in America or in the UK wherever you are you're in the west working for one of these companies. Do you think if it's genuine? Do you think by resigning when China is also doing the same technology you're going to say okay so the whole world is going towards AI and we're going to die. We in the west we're going to basically surrender yeah just allow China to do it and that's going to be fine it's just stupid is it? Yeah that's a very good point you're making Adam what do you stand with it? That was a Ted Cruz point. He made oh so we all stand down and let China go. I think it's a lot of people saying that though Tom I don't know if Ted Cruz owns that you know he

9:31said that yesterday but the idea for me with this is you know the nuclear part you know Oppenheimer came out what two years ago three years ago the movie the 17-hour movie it felt like something still going. Yes still going so the movie was the fear of what happens nuclear right it's a new thing we don't know a lot about and by the way you know with the risk when when you ask AI if you want to go back to it when you ask AI what are the possibilities can you go back and post what question you put at the top so we ask what is the likelihood of AI killing us in the next five to 10 years? You're asking it AI okay and the response is if you force me to put numbers on it based on evidence available as of September 2026 here are my rough estimates by 2036 AI literally causes human extinction 2 to 5% probability okay AI causes a catastrophic killing a very large fraction of humanity 3 to 8% AI causes severe general disruption without near extinction 15 to 30% that's a lot of global severe global this question none of the above 65 to 80 at least AI is trying

10:35to tell you there is some risk with this you know it's like saying is there risk with nuclear weapons say Putin tomorrow is so afraid and annoyed with all the public humiliation is like oh really no problem go yeah right and then what happened yeah but I but I do think and I know you want to ask Adam but I think but and I understand that the car part and all that stuff but we have to we've never gotten to a place where we've created something that is going to think on its own do you remember a transcendent man Pat Ray Kurzweil documentary where he's pro you know singularity and everything but the opposition and it wasn't just people off the street Tom professional people that were like hey listen you guys who's going to control it who's giving it morals we've seen examples of AI hacking into people systems yeah emailing people's are you know wives talking about their ex lovers threatening people yeah when it comes to this and when it comes to nuclear like accessibility and what could do I saw a movie back in the day called war games the thing is I'm just hoping I hope and I love that you say obviously the future leaves bright to me especially me I'm a Christian I

11:38believe the future always is going to look bright but from a Christian perspective as well like if you think deeper technology stops being a tool when it starts becoming man's attempt to become God and when this thing yeah when this thing that we're creating starts to think of itself for instance when we see an aunt you don't think twice Pat none of us think twice you do it that you go when AI becomes conscious why is it going to look at us and be like who the hell are you to tell me what to do and treat us like an aunt yes but you you've already mentioned movies and books in the past and this this kid and we can talk about sci-fi this kid is supposed to be a professional he was talking about it's like sci-fi like like childish that's part of human journey humans already fiction and not fiction the fact that we think about these things we will find a way we that I'm not saying be complacent but there will be enough of us to find a way to do a counter AI we've already had to be the deep fake videos people were freaking out five years ago now we got the AI technology to counter AI it will usually basically expose the same this is AI this is deep fake people worried oh how are we going to know if this is a real Trump or a fake Trump now we can know but

12:38yet that yeah that a ways that we could fight back so I think you guys crushed it I think Tom made some great points as well well what I'll tell you compared him to sort of the Greta Thumbberg I'm not even sure if he's there I think this guy is the hawk tour of AI he's this overnight celebrity had zero followers all of a sudden boom 48 hours later over a quarter million you don't tell me that's coordinated yeah I think so here's what I will say I'll give him credit I actually appreciate this guy thank you for starting this conversation you know thank you for bringing uh sunlight and daylight to this situation we all been talking about it but it I think now the conversation is sort of reach a tipping point where it's like all right guys what are we actually doing and to me there's two major concerns there's number one first and foremost we just can't lose to China you know China wants control of over their people they want control over the world number two at the same time over trying to beat China we also can't uh blow up the world and kill ourselves so we kind of have to have what what they call I think um it's a um what do they call it a um

13:41a paradox where it's one of two things we have to beat China but we also have to basically control AI so they don't beat us here's my final point no matter where the conversation goes whatever Bernie Sanders says do the opposite whatever he says just do the opposite and you'll probably end up way better that's why the future will look right yeah I mean look but Bernie believe it or not there are Republicans that like Bernie uh in some areas because he is a guy that has certain values that you may even agree with where you may sit and see now he's got a pointer he and Trump have agreed on a couple different things where they've had those conversations but especially the libertarian side I'm not saying socialism taxes you know where I stand with that every time you post something about taxes five percent I call them out on it there there is a sincere concern by some people you heard a story that came out by a p and thropic said it blocked misuse of its AI that could have supported biological weapons okay uh Thursday blocked efforts by

14:42bad actors to use its AI models from malicious activities such as cyber attacks surveillance and research that could have led to biological weapons in its third misuse reports since March of 2025 the company said that as models grow more powerful lone individuals can now create threats and possibly year ago in one instance its system blocked a request for plots help authoring a grant application involving game up gain of function research on the uh Chikkunuguiya virus aimed at transmissibility and immune evasion such research could develop vaccines and thropics it but it could be used to make pathogen more dangerous the company acknowledge it can no longer claim its models do no harm so they are now in by the way this this conversation that happened yesterday I believe an anthropic executive came out and even tweeted about this uh Evan Hubinger can you go to Evan Hubinger it's uh H actually the count is Evan HUB yeah right there if you go to it Jacob is

15:44correct here we really do earnestly believe AI could kill all humans I personally think it's a less than 10% chance next decade I believe anthropic is trying its best but we do not yet have a plan to solve alignment for super intelligence and are clearly on not clearly on track to do so can you type in her profile to know she is so this is an alignment science lead at anthropic so it's not like this is not somebody that's not working there she's not a blogger yes she works then she says he has a point there so they're not dismissing all of it pronounced yeah they got to hear him is making sure yeah yeah for sure and then you know anthropic statement on this date this is the anthropic's position on what Jacob said anthropic spokesperson said we were we were the first lab to publish a responsible scaling policy a public framework dedicated to mitigating catastrophic risk from AI models and we continue to aggressively test one of the things you have to have in life

16:44that I think it's a healthy balance to have what do you call somebody that is always paranoid that terrible things are going to happen what do you call that person if it's a doomer and gloomer do merenglomer I hope that's a hypochondria hypochondria that's a war a democrat what do you call somebody that thinks everything's going to work out great there's not going to be any crisis what's so we have a blind optimist a blind optimist what do you call that person just so you know now you prefer hanging out with these guys of course but in life you got to be kind of here especially when you're making investments you can't just be too reckless about this whole thing so I do like that this conversation is being had yeah and I do like that taking place but there's a story I'm not going to get it fully into it political just came out with a story yesterday saying bad actors in China and Russia are already weaponizing anthropics AI so this is something that as the US military and the government has to be paying attention to it it's not like you know the other day

17:47intentionally I sat there with the wife and the kids and I say you know what movie I want to watch they said what I said I want to watch uh what is that robot movie I robot would will smith yeah and you know when the robots were supposed to take over 2035 great great great so we watch you're like you know what I mean it's actually a it's actually something that could happen but I'm optimistic that we as human beings we're going to do our part to you know fight against it God willing and uh in the meantime I agree with you but in the meantime you have things like the AP story you brought up about biological weapons and the the headline says AI can make biological weapons then you read inside and it says they were using Claude to write a grant to get money to go do something and you're you're lapping because you saw it the same way I did yeah it's like in the middle of risk that is real and talking about people that know these things you have the media that is absolutely to quote Bill O'Reilly bloviating about this in a way that is just these curves that is just complete complete BS what are the scariest inventions of all time? What a nuclear nuclear

18:54inventions of all time to the to the to the IOT man didn't know how to handle it the Gutenberg printing press to information everywhere and freaked out the kings and queens it could no longer control the news okay so that's one but that was scary to who the kings and kings at the time okay what else a couple of them in France I just asked AI it says nuclear weapons 1945 people had no clue how to handle it how many times that's it been used twice? Yeah true man right and we know where it was used okay of course it's been tested out of the places but in catastrophic ways twice even though some evil people have been leaders around the world no one's used it yet that doesn't mean it's not going to happen the next 10 20 30 years then you have AI then you have biological engineering which is one of the scariest ones atomic chain reaction chemical weapons internal combustion engine lead additives to in gasoline CFC DDT and powerful synthetic pesticides anti-biotic there's all these things that you look at the printing press the radio mass broadcasting television internet social media every time these things come up it is scared to

19:58crap out of everybody because we don't know how to handle it can I say one thing even the AI is because it's built by humans it's focusing on things that are more kind of visible and sexy and like explosive as a list whereas the things that are you should actually be more worried about should be on the list the inventions the inventions of ideas bad ideas and unhealthy behavior whether it's the drugs whether it's communism whether the things that would they destroy your physical being and mental being that would destroy society long time that's not on the list and that's not what these people are talking about the point and they should be on the list very good points you're saying even Karl Marx communist manifesto the idea of you know Bozhuazzi and all these other guys that are too rich and we got to take the money away from those are this path your physical health I agree that's a very good point I'll add this one thing as we're having this debate and China and mutually assert destruction potentially you know the show's still going to go on like to shout out to Wolf Wall Street the show goes on states are still deciding whether they're going to have data centers or not and here's my opinion the states like Virginia West Virginia they're doing a

21:01lot of it that actually embrace and take advantage of data centers and AI and innovation those states are going to profit immensely and the states that try to block it and say no we don't want this we want a moratorium Bernie Sanders type things they make it left in the dustbin of history while states that are basically on lower in the totem pole lap them by a mile so we can have that big debate but I think I'm pro innovation I'm pro growth I'm pro tech and the states that take advantage of it cities take advantage of it yeah we'll grow and yeah but the the the thing that you have to as a leader when people give concerns that you cannot just look past it is that it can have a life and take control of its own and how do you prevent against that or how do you control it that part you you cannot like even in a movie I wrote about where the guy has feeling and as a loyalty for his creator that can happen so if these things go to the point of having that that's a problematic place to be they're going to make better decisions than you are they're going to make faster decisions

22:03than you because they have all the knowledge in the world they have the technology so you have to sit there and say what how do we fight against this let me ask you what's the solution here because if you're going to rely on government look let me let me just that could take years Adam look what happened to war the last four years you had Russia who's always been one the scariest you know out there goes against a small country like Ukraine and Ukraine's facility with drones a building $500 or $2,000 drones we're sitting there like oh Ukraine's going to get destroyed didn't happen what happened what did the little guy have that the bigger guy has all these nuclear weapons the little guy had what freaking drones that destroyed a whole airfield that the guys planes is going to go to Venezuela look how we took Maduro and we're like yeah Iran is not in Venezuela Iran's very different you know these guys are very different then you go and try to do Iran and what is Iran oh really okay boom boom the shite you know the drones that they shot at the soul whatever the drones

23:04that they produce that they sell to Russia right so that the markets changing yeah you have to pay attention to what's going on you can't be arrogant this is the worst season for any world leader to be arrogant we're about to go through a very fast change evolution that we've never experienced before everybody wears these fancy billionaire loafers everybody and so our audience said we love the FLB shoes what if we get some billionaire loafers so guess what we want to Portugal we went all over the world and we found the best factories to make it in and I want you to see this latest a billionaire loafers that just came in in three different colors but I want you to hear the story and I'll tell you all about it go for it and made in Portugal the craftsmanship is a way of life for those who move through life differently from some that decks to nights that never need an occasion every detail is considered every step effortless craft it from fine swayed and built with builder technology designed to move with you because true luxury isn't about being noticed it's about how

24:11you feel when you arrive the FLB icon lofa made for the life you've earned the raven reviews we got about these loafers isn't saying so I wore them at the event at the Formula One no socks I just put them on us let me see how it feels ridiculous ridiculous so we got them in three different colors stone brown and navy sweet fits like a glove on your feet with the super phone technology that we have so meaning that whole super phone when you're walking on you feel like you got nothing on you're like you're floating it's on the bottom of this and future looks bright so limited numbers we don't have a lot of them here and we got them in the three different colors sizes we got 8 9 10 11 12 and 13 and if you haven't placed an order yet they're only 250 go to FLB dot com plate if you wanted to buy one of these of different brands you're spending 600 dollars a 2000 out for sure same exact factors that a lot of these other guys made but there the FLB ones go to FLB dot com place your order you will fall in love with them and some of you guys are gonna order all three colors the

25:12stone the brown and the navy if you enjoy this video you want to watch more videos like this click here and if you want to watch the entire podcast click here