Skip to main content

Jake Ward

Nov 12, 2026

2:31
So if someone is unaware of everything I'm hinting at right now, tell us... what's uh what's been going on
2:39
yeah lucky you people if you haven't been paying attention to this and if you've been going about your normal week so yeah as of sunday this past weekend jensen huang the ceo of nvidia has been ceo there since 1993 announced to his staff that he had been, quote, unfaithful to my wife and that he'd had an affair.
3:02
And the board announced it was forming an independent investigation around this.
3:08
Now, normally, when CEOs are admitting affairs after the fact... it's bad news for their tenure because by and large, it's not so much the affair that gets CEOs fired, it's having lied to the board about it or in any way concealed it that gets them into some trouble.
3:28
And so that's been true of the CEOs of McDonald's, Nestle, Intel.
3:33
They all lose their jobs.
3:35
The president of CNN, where I work, they all lose their jobs when they have an affair with somebody or an inappropriate relationship with a subordinate and then and conceal it from the board.

5 MINS LATER

9:01
This is
2:37
What's been going on?
2:39
Yeah, lucky you, people, if you haven't been paying attention to this and if you've been going about your normal week.
2:44
So yeah, as of Sunday, this past weekend, Jensen Huang, the CEO of NVIDIA, has been CEO there since 1993. announced to his staff that he had been, quote, unfaithful to my wife and that he'd had an affair.
3:02
And the board announced it was forming an independent investigation around this.
3:08
Now, normally, when CEOs are admitting affairs after the fact... it's bad news for their tenure because by and large, it's not so much the affair that gets CEOs fired, it's having lied to the board about it or in any way concealed it that gets them into some trouble.
3:28
And so that's been true of the CEOs of McDonald's, Nestle, Intel.
3:33
They all lose their jobs.
7:04
What do you mean when you say we knew about
4:09
That's
4:10
what they do.
4:10
Okay.
4:11
Instead, This guy, Evan Hubinger, I've never actually said his name out loud, so I don't know, but he's the AI safety lead, essentially, at Anthropic, writes in a tweet, I agree with Jacob.
4:28
We earnestly believe that there is a greater than 10% chance that what we are building will kill us all within the decade.
4:35
And then he put an exclamation point at the end of it.
4:40
That's a loud exclamation point.

15 MINS LATER

19:16
exactly the skin cancer is that you don't know that's but that's what i'm saying is like it is in a way deadening the signal a little bit or anesthetizing the signal a little bit and that worries me
2:20
And so let me let you, Jake, introduce this topic to us.
2:25
All right.
2:26
So on Tuesday, September 8th, I was gearing up for what I thought was going to be a pretty quiet day.
2:32
And then this freaking guy, Jacob Coxon, a pre-training researcher at Anthropic Research, He's been in the field for about three years, which makes him a veteran of the frontier model world.
2:46
Decides to publicly denounce his resignation from not only Anthropic, but from the whole field of AI because of what he said was incredibly irresponsible behavior on the part of both of his prior employers, both Anthropic and OpenAI, where he spent three years and contributed meaningfully to the creation of one of their iterations of ChatGPT.
3:06
He basically said, I'm quitting this field.
3:08
I'm retiring from, you know, I'm getting out of this job because you can't trust either of these companies.

16 MINS LATER

19:26
worries me
2:20
And so let me let you, Jake, introduce this topic to us.
2:25
All right.
2:26
So on Tuesday, September 8th, I was gearing up for what I thought was going to be a pretty quiet day.
2:32
And then this freaking guy, Jacob Coxon, a pre-training researcher at Anthropic Research, He's been in the field for about three years, which makes him a veteran of the frontier model world, decides to publicly denounce his resignation from not only Anthropic, but from the whole field of AI because of what he said was incredibly irresponsible behavior on the part of both of his prior employers, both Anthropic and OpenAI, where he spent three years and contributed meaningfully to the creation of one of their iterations of TataGPT.
3:06
He basically said, I'm quitting this field.
3:08
I'm retiring from, you know, I'm getting out of this job because you can't trust either of these companies.
3:14
And he said that they are gambling with our lives.

16 MINS LATER

19:16
exactly the skin cancer is that you don't know that's but that's what i'm saying is like it is in a way deadening the signal a little bit or anesthetizing the signal a little bit and that worries me
5:40
Yes.
5:41
He may never get a job in this industry again.
5:42
And the fact that he's willing to blow up his stake in this industry to say this, I think is part of what caught everyone's attention.
5:50
So I got calls, multiple ca- So I have a contributor deal at CNN, and I got multiple calls from CNN, s- different producers on different shows saying, "Do you know this guy? Can we get this guy?" And finally Anderson Cooper did get him- Uh, last night and, and had him on.
6:05
And yeah, he just says, yeah, I just think this thing is, is improving itself and that the industry is incapable of slowing itself down.
6:13
He also said that he believes that the industry is earnestly interested in being regulated because they can't slow themselves down.
6:19
So there's this whole, like, arms locked together running toward a cliff, and no one can agree on how to slow down or whether to slow down.
11:09
Mm-hmm.
47:01
So what have you used a 3D printer for in your home?
47:04
When I bought my first 3D printer, my wife was extremely skeptical.
47:08
I decided to kinda look around the house and see if there was anything I could print to make her life a bit easier.
47:13
The first thing I printed was our toddler has these bibs that she'll use when she eats, and we didn't have a great way of drying them.
47:20
So I printed this custom double hook that came down from our cabinets, and we could hang them right there above the sink, and it would dry perfectly.
47:28
And the next thing I printed was she would go on these long walks with the stroller, and there was nowhere to put her large water tumbler.
48:18
And so are there any limitations to 3D printing?
48:22
One big one is the size.
41:26
Can you just talk about what, what the most damning or, or impactful evidence has been thus far?
41:32
Well, the hardest thing, Anderson, with this, right, has been being outside a company as secretive as Meta, and all tech companies keep their secrets close.
41:41
But now because of the discovery that this trial makes possible, we're seeing all kinds of internal research, and we're hearing from internal people.
41:49
So the star witness for the states is a guy named Arturo Bejar, who basically ran Facebook Safety for a time, then worked on wellbeing at, at, uh, Instagram.
41:57
He met with Zuckerberg, he says, over 100 times.
42:00
And he says essentially that Meta was measuring harm pretty accurately inside its walls, but publicly giving us very different numbers, which is pretty much at the heart of what the states are saying here.
42:12
He wrote a long letter, sort of a desperate final plea as he put it, to Mark Zuckerberg on the day that Zuckerberg was responding to a big whistleblower, uh, situation.
42:45
What did he say?
42:12
Can you explain in terms that I can understand, that's a tall order, uh, how AI was used in this experiment to create the viruses?
42:20
Well, John, right, what we've been trying to figure out for a long time is how, for instance, would you create new antibiotics, right? That's a huge problem that we have in the medical field right now.
42:31
We, we have all kinds of bacterial infections you can catch out of hospitals that current medicines really cannot fight.
42:37
And part of the problem is the incredible trial and error it takes to figure out what materials in nature might be the basis of a new antibiotic.
42:45
In this case, what you've got is basically all of that trial and error collapsed into the kind of efficiency that AI, you know, is always sold on.
42:55
And so we have these scientists who might have spent, you know, years, a lifetime trying to figure out are there any viruses in nature that could conceivably kill these bacteria in a Petri dish.
43:05
Instead, they could use AI, pull in some data, build 300 or so new viruses.
43:38
Mm-hmm.
4:05
what I mean? Yes.
4:06
And there's that vibe to the language that OpenAI used in describing the thing that this model did as unprecedented behavior.
4:17
You know, they like to describe it as like an Ocean's Eleven kind of heist.
4:21
You know, they're just obviously like excited about this as much as they are embarrassed, if they are in fact embarrassed about it, which I just think is really interesting.
4:32
And it seems to have created, and jump in here anytime I can, but it seems to have created now...
4:38
There's like a lot of simultaneous stuff happening that is creating this kind of question of, first of all, have these models gone too far? And second of all, should these companies now be pulling back? Because one of the things that several observers pointed out is that OpenAI actually has an internal framework for evaluating this stuff.
5:01
And according to that framework, depending on how you read it, once a model gets to a place where it can hack basically other things, things without human intervention.

17 MINS LATER

22:02
And so I am looking forward to talking to you about this, Jake, because I would like to hear, you know, what did you ever have Winamp running on your machine? Oh, yeah,
2:43
Yeah, it's been wild.
2:44
Yeah, of course you have.
2:45
Of course, wild, right? So to catch anybody up who hasn't been paying attention to this, uh, OpenAI's...
2:51
A, a pair of OpenAI's models were sicked on the task of, of doing some, some cybersecurity work, which as we all know is code for hacking, and, uh, they were given instructions, left in a sort of sandbox, as far as I can tell unattended, and wound up, uh, going out onto the open web, finding a way out onto the open web, and breaking into this kind of beloved shared library company called Hugging Face, where they were looking for the answer key to this particular problem that they'd been given.
3:24
And the crazy part is Hugging Face found this out, whatever this was, almost two weeks ago now, and alerted the FBI, 'cause they knew that they had been attacked, right? But didn't know that it was OpenAI's product that had done it, and in fact, according to people I've talked to at Hugging Face, OpenAI didn't know it was their product until very recently.
5:56
I hate that.
5:56
Yeah, I don't like that.
5:58
And so there is a really interesting- ... thing going on there.
2:44
[laughs]
2:44
Of course, wild, right? So to catch anybody up who hasn't been paying attention to this, uh, OpenAI's...
2:51
A, a pair of OpenAI's models were sicced on the task of, of doing some, some cybersecurity work, which, as we all know, is code for hacking, and, uh, they were given instructions, left in a sort of sandbox, as far as I can tell unattended, and wound up, uh, going out onto the open web, finding a way out onto the open web, and breaking into this kind of beloved shared library company called Hugging Face, where they were looking for the answer key to this particular problem that they'd been given.
3:24
And the crazy part is, Hugging Face found this out, whatever this was, almost two weeks ago now, and alerted the FBI, 'cause they knew that they had been attacked, right? But didn't know that it was OpenAI's product that had done it, and in fact, according to people I've talked to at Hugging Face, OpenAI didn't know it was their product until very recently.
5:56
I hate it.
5:56
Yeah, I don't like that.
5:58
And so there is a really interesting- ... thing going on there.
6:02
And then sort of tangential to that but, but is sort of interesting is Nvidia then on, uh, let's see, when was that? Mm, three days ago, Monday, launched this new, what they're calling the Open Secure AI Alliance with this long list of companies.
58:29
[laughs]
58:29
But yeah, just doing that, and, you know, it, it's literally just look at a video, "Yeah, that's, that's quite funny." Look at it, go back to the start, call it, chuck it on social media.
58:40
The other week I, I was calling, you know, a bloke having a shot for goal.
58:43
Kicks it, hits a seagull.
58:45
Seagull gone, over.
58:47
So, um, g- going back though, I will, um, absolutely any sort of, any sort of footy, uh, I'm all for it, I, I think.
58:57
A- and I don't- I'm not really good at many things, but just talking absolute nonsense for two and a half hours while the game's going on.
63:41
It's chaos
12:14
But how would it change? Can we have a peek or preview into how it'll change? I'm trying to explain from my knowledge of a development
12:20
developmentWell, I think, I think you nailed it with that.
12:22
You've changed the floor, right? A, a developer can replace dozens, hundreds of hours by prompting a, a code IDE that supports their language, right? I want to achieve this goal, here are the variables, here are the parameters, write the base code.
12:42
That's incredibly efficient.
12:45
It also removes the need to, to have perfect code structure, right? Like, you have, you have enabled a machine to do a commodified job that for a long time we had heads and hands coders developing, that we would, um...
13:00
You know, some people derogatory, a derogatory name for them is code monkeys, right? Like somebody that just bangs on code and it's heads and hands and I'm moving this stuff around.
13:10
But they were, they were writing lines of code to meet a demand because software development is about predicting the future.

10 MINS LATER

23:20
Mm-hmm
0:27
I know it's an exaggeration, but what's your view on that as an association of, uh, developers? What do you see going forward? How this developer industry or developer role shifting, changing?
0:40
Yeah, I, I think reports of the demise of the developer ecosystem have been greatly exaggerated.
0:45
I, I think that, uh, I suspect that what we're going to see in the ne- So I saw that...
0:52
I s- I saw a couple weeks ago, there was an article that I, I quite enjoyed about how AI has actually generated an entire generation of decoders that are actually coding again, that the no-code, low-code movement over the last three years where it's so easy anybody can do it, was actually removing some of the creativity that goes into the development of software as a rule.
1:14
But we're 15, 16 years out from the idea of software eating the world.
1:19
It ate the world.
1:20
It's gone.
4:48
But how would it change? Can we have a peek or preview into how it'll change? I'm trying to explain from my knowledge of development.
1:45
Tell us a little bit more about the Developer Alliance and about your data protocols business.
1:50
Sure.
1:50
So in 2012, I co-founded the Application Developers Alliance.
1:54
It feels like a million years ago now, though many of the same issues we dealt with then we deal with now.
2:00
I started the organization as an alliance of companies that invested in software developers that depended on them as both their workforce, but also their partners.
2:07
So you're talking about a lot of platforms, you know, your Googles, your Metas, your...
2:11
Intel's IBM, et cetera.

33 MINS LATER

35:11
So we're going to do it anyway, that sort of thing.
24:50
[laughs]
24:50
So, you know, you usually have one or two in a booth, maybe a third.
24:55
You know, our sport's just trying to find its feet of who's actually good and, and who actually isn't.
25:00
So, uh, the world love the Australian accent, the bit of the banter, and I've just been lucky enough, right place, right time, and have spent the last three, four years traveling the world, uh, seeing a lot better matches between you and Howie.
28:03
Right.
28:03
But we just had a guy drop, I think, 27 kilos for the match.
28:08
He did it naturally and then dropped eight or nine kilos.
28:10
So you drink copious amounts of water, um, during the week for a very long time, and then on game day, you just sweat it all out, and it just, uh, does something to the body.
32:27
Tell us about what this encyclical is and why our latest Pope Leo XIV decide what, what, excuse me, why did Pope Leo decide artificial intelligence was even worth an encyclical?
32:41
Okay.
32:41
So I love this.
32:42
So, so I should just say here, first of all, I am not Catholic.
32:46
I am not a Catholic.
32:47
I'm not a Christian.
32:48
I'm not particularly religious.

9 MINS LATER

41:42
What do you think that says, though, about Anthropic kind of being the headline there?
35:31
[laughs]
35:31
Like, this guy, w- he was in Peru under the sh- dealing with the Shining Path as a young priest.
35:37
He did like 10 cushy years in Rome, and then he went back to Peru where he's, like, driving vans through floodwaters and dealing with some really scary situations.
35:46
Like, this guy is n- is n- no joke.
35:48
He's not some, you know, uh, pencil neck intellectual.
35:52
He is a very serious person, uh, out in the world and a very s- smart intellectual about this stuff.
35:59
And the reason he's called Pope Leo is he named himself a- after Pope Leo XIII, who in 1891 was the first guy to say, "Factories are a nightmare and are creating nightmarish conditions for children who are being forced to work in them 14 hours a day." And that was the first time that the Catholic Church had ever really stepped into social concerns.

13 MINS LATER

48:52
Yes
34:31
[laughs]
34:31
Like, this guy, w- he was in Peru under the sh- dealing with The Shining Path as a young priest.
34:37
He did like 10 cushy years in Rome, and then he went back to Peru, where he's, like, driving vans through floodwaters and dealing with some really scary situations.
34:46
Like, this guy is n- is n- no joke.
34:48
He's not some, you know, uh, pencil neck intellectual.
34:52
He is a very serious person, uh, out in the world and a very s- smart intellectual about this stuff.
34:58
And the reason he's called Pope Leo is he named himself a- after Pope Leo XIII, who in 1891 was the first guy to say, "Factories are a nightmare and are creating nightmarish conditions for children who are being forced to work in them 14 hours a day." And that was the first time that the Catholic Church had ever really stepped into social concerns.

13 MINS LATER

47:52
Yes
7:50
Can you talk about that as well? Sure.
7:52
Yeah, so it's been a really interesting parade of characters.
7:57
And as a result, it's been a really amazing tour of...
8:03
the sort of social realm that these people occupy the weird sort of background assumptions like the sort of underlying assumptions of their lives that we can get into and one of those is that they should just have like unimaginable amounts of money and yet somehow be above money is a weird thing that keeps coming up um And then, as you say, the accusation levied against Altman by many of the witnesses, including people that he founded the company with, including people he has worked with for years, board members, all of them saying, this guy's a liar.
8:43
He lies and lies and lies.
8:45
And so I'll just give you sort of a rundown of a couple of the sort of most interesting character stuff that has sort of jumped out at me.
8:51
So one really core person that I think was really the sort of heart and soul of this is Ilya Sutskever.

12 MINS LATER

20:51
And like, is there shouting that takes place at any time? Is there a lawyer whispering in someone's ear? No,
1:57
[laughs]
1:58
Um, so Elon Musk is suing Sam Altman, Greg Brockman, the president of OpenAI, and OpenAI itself, and kind of by extension, Microsoft, I guess, for converting from a nonprofit to a for-profit, although it's a B corp, so kind of whatever you wanna call that, a hippie, hippie profit.
2:20
Um, and the question is essentially did Mu- sorry, did Altman and Brockman trick Musk into donating a bunch of money to get OpenAI off the ground, and then pull a bait and switch on him, kick him out, and make it into a for-profit company that's now poised to eat the world? And what Altman's side is arguing is, um, Musk is a, m- you know, a, a megalomaniacal control freak who wanted, uh, total control over OpenAI when he and Altman began it together in 2015, and when he was not given that amount of control, he walked away, he took his football and went home, and, uh, and now that he has seen OpenAI succeed, where xAI, uh, Musk's AI company, has fallen so far behind, he's now suing to try and slow them down.
3:25
So that's the two... those are the two sides of it, and the implications of this are, right, the stakes here is that what he's asking for is the money back that he put in, which doesn't really matter 'cause money is-- comes up out of the ground for these guys.

5 MINS LATER

8:59
Mm.
8:59
So Ilya Sutskever is one of the co-founders of the company.
9:02
He, at the age of twenty-four, back in 2012, uh, wrote a paper when he was at the University of Toronto, um, about, uh, an image recognition software system that he'd created, um, with Geoffrey Hinton and this other guy, I can't remember the third guy's name.
9:16
His name is Alex something.
1:57
[laughs]
1:58
Um, so Elon Musk is suing Sam Altman, Greg Brockman, the president of OpenAI, and OpenAI itself, and kind of by extension Microsoft, I guess, for converting from a nonprofit to a for-profit, although it's a B corp, so kind of whatever you wanna call that, a hippie, hippie profit.
2:20
Um, and the question is essentially did Mu- sorry, did Altman and Brockman trick Musk into donating a bunch of money to get OpenAI off the ground, and then pull a bait and switch on him, kick him out, and make it into a for-profit company that's now poised to eat the world? And what Altman's side is arguing is, um, Musk is a, m- you know, a, a megalomaniacal control freak who wanted, uh, total control over OpenAI when he and Altman began it together in 2015.
3:02
And when he was not given that amount of control, he walked away, he took his football and went home.
3:12
And, uh, and now that he has seen OpenAI succeed where xAI, uh, Musk's AI company, has fallen so far behind, he's now suing to try and slow them down.
3:25
So that's the two, those are the two sides of it.
3:27
And the implications of this are, right, the stakes here is that what he's asking for is the money back that he put in, which doesn't really matter 'cause money is, comes up out of the ground for these guys.

5 MINS LATER

8:58
Mm.

We value your privacy

We use cookies to understand how you use our platform and to improve your experience. Click “Accept All” to consent, or “Decline non-essential” to opt out of non-essential cookies. Read our Privacy Policy.