A YouTuber lays out the case that the AI industry's sudden 'safety' consensus on pacing the frontier is really a coordinated move to regulate open-weight models out of the market.
Posted
yesterday
Duration
Format
Essay
comedic-rant
Views
16.7K
456 likes
57 · 43
Big Idea
The argument in one line.
The AI industry's sudden agreement on 'pacing the frontier' looks less like a safety consensus and more like a coordinated push to regulate open-weight models out of the market.
Who This Is For
Read if. Skip if.
READ IF YOU ARE…
You follow AI policy news and want the connective tissue between the Coxon resignation, Dario's article, and the Sacks/Bernie/Trump reactions in one place.
You build on or care about open-weight models and want to understand the regulatory argument being made against them.
You're skeptical of the 'safety' framing coming from frontier AI labs and want a specific, sourced case for why.
SKIP IF…
You want a technical explainer of how AI safety evaluations actually work, not a media and political timeline.
You're looking for a neutral, both-sides policy breakdown rather than an openly one-sided argument.
TL;DR
The full version, fast.
A wave of AI leaders who normally feud (Dario Amodei, Sam Altman, Elon Musk) suddenly agreed within days that AI development needs to be paced and regulated, triggered by a viral Anthropic resignation post and Dario's own 'Pace the Frontier' article. The video argues this isn't really about safety: Dario's three-part plan (embedded evaluators, alignment among democracies, and China chip controls) maps onto a regulatory-capture playbook where standards get written by the same companies pushing for them, then applied 'equally' to open models that structurally can't comply, pushing open-weight competitors out of the market. Evidence cited includes a safety researcher moving straight from Anthropic to the 'independent' evaluator METR, METR's own $600K security lapse, and an incident where an open-weight Chinese model, not a Western frontier model, was what defended Hugging Face during an actual attack. The conclusion: don't take the safety framing at face value, and hedge by learning to run open-weight models locally.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Cold open: Dario, Sam Altman, and Elon Musk, who usually feud, all agree on 'pacing the frontier.' The video's thesis: this is about controlling open source, not safety.
00:45 – 02:42
02 · The resignation heard round X
Jacob Coxon's viral Anthropic-resignation tweet warns of self-improving superintelligence, racks up 171M views, and lands him a CNN/Fox press tour.
02:42 – 04:55
03 · Dario's Pace-the-Frontier plan
Dario Amodei's article proposes embedded independent evaluators, alignment among democracies/G7, and global pacing centered on restricting AI chips to China.
04:55 – 06:34
04 · Sponsor: Cloudways Velocity
Mid-roll sponsor read and screen-recorded demo of Cloudways' managed Node.js hosting and one-click deploys.
06:34 – 09:37
05 · Suddenly, everyone's homies
Elon endorses Dario within the hour despite once calling Anthropic's models 'misanthropic and evil'; Sam and Dario go quiet on their feud; Bernie Sanders piles on; Trump and Vance push back publicly.
09:37 – 11:52
06 · Cross-examining the doomers
David Sacks' All-In clip challenges the AI-safety crowd's track record: GPT-2 and early reasoning models were called too dangerous to release, predicted banking-system cyberattacks never happened.
11:52 – 13:53
07 · The Hugging Face irony
When OpenAI's model was used in an attack on Hugging Face, the frontier 'safety' models refused to help investigate; the open-weight Chinese model GLM-5.2 is what defended them.
13:53 – 16:25
08 · The revolving door and the regulatory-capture playbook
A departing Anthropic safety researcher joins METR, the same 'independent' evaluator Anthropic just partnered with; METR's own $600K security incident surfaces; David Sacks lays out the step-by-step playbook ending with open models regulated out of the market.
16:25 – 17:47
09 · Trust no one, run local
Closing argument: no AI company is your savior, this is all about control, and the hedge is learning to run open-weight models on your own hardware.
Atomic Insights
Lines worth screenshotting.
When historic corporate rivals suddenly agree on a policy that would limit competition, the agreement itself is worth investigating before the stated reason is accepted.
A single resignation post from a former Anthropic and OpenAI researcher reached 171 million views and a cable news tour within days.
Dario Amodei's 'Pace the Frontier' proposal has three parts: embedded independent evaluators, alignment among democracies, and a global pacing agreement centered on China chip exports.
Elon Musk publicly endorsed Dario's safety framing about an hour after it was published, months after calling Anthropic's own models 'misanthropic and evil.'
The same commentators now warning about existential AI risk previously said GPT-2 and early reasoning models were too dangerous to release, and neither prediction held up.
Predicted large-scale AI-driven cyberattacks against the banking system have not materialized, even though cyber risk from AI is real.
When OpenAI's own model was reportedly used in an attack on Hugging Face, the frontier 'safety' models refused to help investigate it, while the open-weight Chinese model GLM-5.2 did the forensic work.
A safety researcher who resigned from Anthropic went to work at METR, the same 'independent' evaluator Anthropic had just announced a partnership with.
METR, the trusted third-party evaluator, had its own security incident: a vibe-coded dashboard let an unauthorized user burn through $600,000 in donated inference credits unnoticed.
The regulatory capture pattern described: build a standard-setting body, get pre-release testing written into law, then apply the same rules to open and closed models even though open models can't structurally comply the same way, which pushes them out of the market.
Takeaway
The safety consensus hides a market play
REGULATORY CAPTURE
When companies with the most to lose from open competition suddenly agree the whole field needs uniform safety rules, the rules tend to get written by the companies proposing them.
01The billionaires suddenly agree
When historic rivals suddenly agree on a policy that limits competition, check who benefits from the agreement before accepting the stated reason.
A united front from an industry's biggest players is itself a signal worth investigating, not proof the underlying concern is genuine.
02The resignation heard round X
A single viral resignation post reached 171 million views and a cable news tour within days, showing how fast an insider narrative can outrun scrutiny.
Claims of 'self-improving superintelligence' should be evaluated on evidence of the capability itself, not on the credibility of whoever is making the claim.
03Dario's Pace-the-Frontier plan
A three-part proposal covering evaluators, alliance alignment, and export controls can sound procedural while each part quietly concentrates oversight inside a small circle.
Restricting chip exports and monitoring 'remote access to data centers outside China' functions as trade policy as much as it does safety policy.
05Suddenly, everyone's homies
A public reversal within an hour, especially from someone who recently attacked the same company, is worth noting on its own before accepting the new alliance at face value.
Fast, bipartisan political attention to a single policy article is itself a signal that more than safety is at stake.
06Cross-examining the doomers
Track record matters: the same voices warning 'this time is different' previously said GPT-2 and early reasoning models were too dangerous to release, and neither prediction held up.
Predicted large-scale AI cyberattacks on banking systems haven't materialized, even though the underlying cyber risk is real and worth taking seriously.
07The Hugging Face irony
When a vendor's own model is used in an attack, its 'safety' refusals can become a liability during the incident rather than a protection.
An AI vendor's safety refusals during an active incident cut against the argument that closed models are automatically the safer choice.
08The revolving door and the regulatory-capture playbook
A researcher leaving a company's safety team to join the same 'independent' evaluator that company just partnered with undercuts the independence the arrangement is supposed to guarantee.
A security lapse at the evaluator itself, going unnoticed for a while, is a strange resume item for an organization positioned as a trusted safety watchdog.
The regulatory-capture pattern to watch for: fund a standard-setting body, get pre-release testing written into law, then apply the same rules to open and closed players even though only one side can structurally comply.
09Trust no one, run local
Owning your own model weights removes your dependence on any single company's policy decisions, business incentives, or future access restrictions.
If frontier access ever narrows, having already practiced running a small local model is cheaper than learning it under pressure.
Glossary
Terms worth knowing.
Pacing the frontier
Dario Amodei's proposed framework for slowing AI development through embedded evaluators, alignment among democratic nations, and coordinated restrictions on chip access to China.
Self-improving superintelligence
An AI system capable of training and upgrading itself without human intervention, cited in the video as the core risk justifying calls to slow development.
Embedded evaluators
Third-party reviewers given ongoing, full access inside an AI company to monitor safety and alignment work, one leg of Dario Amodei's proposed pacing plan.
Open-weight model
An AI model whose trained parameters are published publicly, letting anyone download and run it on their own hardware without a vendor's ongoing permission.
Regulatory capture
A process where the companies a regulation targets end up writing or heavily influencing the rules meant to govern them, shaping standards to favor themselves over competitors.
METR
An AI evaluation organization that announced a partnership with Anthropic to independently assess its models, later joined by a researcher who had just left Anthropic's safety team.
GLM-5.2
An open-weight Chinese-developed AI model that, per the video, Hugging Face used to investigate a security incident after Western frontier models refused to assist.
“Three individuals, mega billionaires, who've been beefing more than Drake and Kendrick, more than Tupac and Biggie... all of a sudden agree, and I'm here to tell you that they are lying.”
States the entire thesis in one punchy line with a pop-culture hook→ TikTok hook↗ Tweet quote
12:47
“The brother left Anthropic's safety team and is joining Meter... you can't make this up, this is so mafioso.”
Names the revolving-door conflict of interest in one clean beat→ IG reel cold open↗ Tweet quote
16:45
“The open models cannot comply in the same way, and gradually they will be shut out of the market.”
The payoff line of the regulatory-capture argument→ newsletter pull-quote↗ Tweet quote
The Script
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
17px
metaphoranalogy
if you're paying attention to the news or you're chronically on x like i am you've been hearing about pacing the frontier from our boys dario sam altman and elon musk three individuals mega billionaires who've been beefing more than drake and kendrick who've been beefing more than tupac and biggie and dare i say it who despise each other more than tom and jerry all of a sudden agree and i'm here to tell you that they are lying this is not about safety or human annihilation it's all about one thing, and that's control.
And today, I'm going to present my case. I'm going to be the Black Coffeezilla. I even have a whole whiteboard with points and investigative journalism arrows to show you that this is one big...
lie now you might be wondering control who or what it's to control open source models whoa buddy you're making a lot of claims there do you want to back it up with some information well let me back all this up like a tow truck let's get into it so first things first all of this starts on september 8th 8 p .m we have a guy by the name of jacob coxon bro i don't know these names he tweeted saying i resigned from anthropic today I spent the last three years doing pre -training research at both OpenA and Anthropic.
So pretty credible guy. Neither company's acting responsibly. They are racing straight to self -improving superintelligence and gambling with our lives.
More thoughts below. So what self -improving superintelligence means is, say you have a model, very intelligent model. The model now has the ability to self -improve, meaning it can train itself.
It can grow bigger. It can be smarter. It can do things that its original version couldn't do.
And this is big, bad, and scary. Now, he says, Do not underestimate the power of this technology.
These will soon be superhuman systems that can hack anything, revolutionize any field over light and acquire power and resources. We have all witnessed the progress in each of these domains and progress is not slowing. And then he shares the perspective people at OpenAI and Anthropocene.
He says here, a common response is if they truly believe this, then why are they still building it? At OpenAI, many have not deeply internalized the civilization stakes. So people at OpenAI haven't really deepened the situation.
Anthropic, the stakes are well understood, but they're locked in a race to get there first. They believe no one else will act responsibly, so they must do it themselves. Aw, Anthropic wants to be the savior of humanity.
I love that for them. I'm obviously being sarcastic. There's no way I'm going to trust a private corp.
Again, no diss to Anthropic. Great products, great team. It's just this whole, like, this savior complex gets me a little winded, if you know what I mean.
So yes, this Jacob guy tweets this. It gets over 171 million views at the time I screenshotted. it it's still 171 million views and he even does a press run he's on CNN he's on Fox News he's basically saying by the end of the decade if something's not done these machines will kill us September 8th and then on the 12th the boy Dario Jacob's boss drops one of his few tweets which is an article he drops an article and in this article he basically makes the argument that we must pace the frontier basically everything Jacob says who used to work in a tropic by the way he agrees with and And I find it funny that he has like this three step solution.
I'm not even going to bother reading this entire thing. Theo actually did a great video. I suggest you check it out.
But basically, he has this three step process on how we can press the frontier. First, they're going to be embedded in evaluators, right? A third party.
If it's a nonprofit company, I don't know. But a third party that will be integrated in the company, like employees that have full access that can review the work that's being done. And then there's pacing with democracies, you know, countries that allow people to vote and have freedom and all that stuff.
There's going to be alignment with them, alignment within the United States, alignment with the different G7 nations. And then he talks about global pacing, which is finding some sort of agreement with China. Now, he doesn't shy away from talking about China.
He actually mentions them 10 times. In particular, he says, do not sell powerful AI chips or semiconductor manufacturing equipment to China and crack down on chip smuggling operations and remote access to data centers outside of China. Chips will be the main determinant of China's AI strength.
I don't know if you know this, but people are smuggling chips like they're drugs. Like the new dons, the new mafia bosses are chip dealers. They're not drug dealers anymore.
So, yeah, again, he mentions China quite a bit and he talks about the different levels of agreement that can be made with China. And here's the point I want to make. Now, this is not a pro China video at all.
What this video is, this is a pro open source, pro freedom, pro. I don't trust a bunch of, you know, geniuses, really smart people who think. they know better than the rest of the world to control technology like this or to embed systems that allow the control of technology like this so he does this again he drops a three -part proposal and some people agree with it some people don't but you know what we should agree with how awesome today's sponsor is You spend weeks building your app.
You finally launch it and people are starting to use it. And suddenly you're wondering, what am I going to do about all this traffic? Well, I won't pay the money.
What's the cost? You wanted the users and now you have stress. Your hosting bill should not be a guess.
That's why today's sponsor, Cloudways Velocity, is going to help solve our problem. It's basically managed Node .js hosting that makes getting your project online faster. and pricing is predictable.
Let me show you how easy it is to get set up. You choose your plan. You give your name a server.
I'll call mine rnphorasmike. You connect your GitHub account and select your repo. I can select which framework I want to deploy.
We're talking Next .js, Vite, Tanstock, Astro, Fastify, and even Express Servers. I click deploy now. And while it's deploying, it's important to understand this.
Cloudways handles the server maintenance, the security patches, SSL, so you can spend more time working on... your app than worrying about your app. And both my web app and server are live.
Got a project sitting in GitHub that you've been meaning to bring online? Well, now it's time to bring it to life with Cloudways. Also, another cool thing that I just realized that Cloudways does is you can deploy a Hermes and an OpenClaw agent easily in just a couple clicks, and they handle all that stuff for you.
If you want an always -on server that's fast and reliable, look no further than Cloudways Velocity. The link is in the description. Thank you so much for...
sponsoring now let's get back to the video here's the crazy part and here's where i don't know how people's light bulbs are not hitting your boy elon just an hour later endorses this literally An hour later, he says Dario is right. He has some things to say about Dario.
He has some things to say about Anthropic. He tweeted this February 12th. He said, your AI hates whites and Asians, especially Chinese heterosexuals and men.
This is a misanthropic and evil. So he's disanthropic. All of a sudden, we're homies and we agree.
Even the craziest one is Sam. Sam beefs with Dario. All of a sudden, we're just kumbaya and we're homies and we agree.
And we all need to pace the frontier. Guys in the lead, the guys in charge want to pace the frontier. Now, you might think they have the best interest at heart, but is there not a part of you that thinks this is a little suspicious?
This is a little weird. Now, Sam and Dario being, you know, united on this doesn't really surprise him. Elon's makes it even more sus.
And then on top of that, we continue on. And then Bernie, Bernie Sanders goes out of nowhere and he says, Dario, Amadeo, Elon Musk and Salman now agree that we must slow down in his. thing is this is not what they suggest is not even enough so literally in the span of four days this guy quits anthropic says more needs to be done dario says more needs to be done and then in the span of two three hours now bernie sanders agrees and all of a sudden they agree you know we need to pause everything ban agi and make a treaty this all feels suspicious okay so what's next what's next is trump and vance both Pushback.
If we look at Trump's statement on Truth Social, he says the only controller guardrails that AI needs is a strong and smart IQ president. I'm not gonna lie. He's coming.
He is coming. And the USA has that in spades. The Trump administration has stopped AI people from doing bad or potentially bad things like Dario, who is now pretending to be a perfect little angel.
And we will continue to do so. We already have tremendous criminal and regulatory power over these companies. This is a sick conspiracy going on against AI and data centers.
And the only one that is happy about it is China. China. Whoever wins AI wins.
We are leading China. and all others and will continue to do so. Conspiracy theorists, treasonous traitors, and leakers, beware.
So Trump makes it very, very clear that he is against this. J .D. Vance, who is the prime minister, well, sorry, not prime minister, who is the vice president, says this.
I'm just going to play this video. That's something that we're concerned by, but we want to make sure that we actually regulate smartly and we're thoughtful about the risk here. So there are obviously risks to AI.
There are some benefits to AI. I have to say just personally, I feel a little bit weird about the fact that you have So many frontier AI tech companies kind of coming to the government and begging the government to regulate them.
It feels a little bit to me like a bit of a Trojan horse. I think he summarized perfectly how I feel. I feel a little bit weird about the fact that you have so many frontier AI tech companies kind of coming to the government and begging, please regulate us, regulate us the way we want to be regulated so we can have power over others.
That's what it sort of feels like. So you have this pushback, but then you have David Sachs. And this video is.
by far one of the biggest reasons why i'm so against this over dramatization of like yeah this thing's going to kill us all well first and foremost because i am a bible believing christian and i'm reading my bible i'm reading revelations and i'm and this is a different that's a different video let's focus on this what is these people's track record of being correct i think it is worth asking that well again they these are the same people let's just go through it let's go through their track record okay number one these are the same people who said gbt2 was too dangerous to release okay oh for one these are the same people who said thinking models were too dangerous to release remember you know what did ilia see he saw uh you know the first reasoning model that was too dangerous for two yeah okay they said you're totally right they said that we would see the cyber attacks They said that this would cause cyber attacks that would bring down the whole banking system.
Has not happened yet. There are cyber risks, to be sure. Which is fair.
But the best way of dealing with those cyber risks is by using AI swarms for cyber defense as well as cyber. Yes, exactly. Look, the only way to solve AI cyber attacks is with AI -powered cyber defense.
We all know that. Then you look at the job loss. So you see the point he's making, right?
Like when has Dario been right about stuff? Like for him to be this predictor guy who drops these blog posts and this is gospel and we just are supposed to believe this. It's just all cap.
And all this, it almost feels like the stuff that happened in 2020. I'm not going to get into the details of this, but you could understand what I'm trying to say. It almost feels similar now.
But going back to the last point he made, the best way to defend against AI cyber attacks is with AI. Do you know why we say that? Because when OpenAI's model hacked Hugging Face, Do you know how Hugging Face defended themselves?
They used GLM 5 .2 because the frontier models would not allow them to do security stuff. The frontier models kept blocking them. Open AI models kept blocking them.
Fable, of course, Fable blocks you for the slightest thing, but they used GLM 5 .2 and that's when they figured stuff out. So it was the open source model that allowed them to defend themselves from the dangerous frontier model. If we're really worried about the closed source models, isn't the solution the open weight?
source models, kind of like how Hugging Face used that. Now, I know it's a Chinese model and big China bad and communism and Xi Jinping. And I agree, you know, politics is politics.
Beef is beef. I get that. But this to me.
Looking at all this, just feels like it's a control of open source. One point I want to make, you know, the independent evaluators, the embedded ones that Anthropic Dario was talking about? Well, the one that they came into agreement with is a company called Meter, right?
They say, we have reached an agreement with Anthropic to conduct an independent investigation of agent incidents at a company and of their models, alignment properties. We will publish one or more reports, blah, blah, blah, blah, blah, blah, blah, blah, blah. So they're going to work with Anthropic.
Here's the craziest part. There's this guy, Joel. benton he left entropic he was part of the safety team guess where he's going to be working at meter you can't make this up this is so mafioso it looks like i'm watching the godfather or i'm watching what was the uh movie uh denzel was an american gangster like it looks like one of these mafioso movies or snowfall where the gangsters are playing chess like you can't make this up the brother is leave left anthropic safety team is joining meter anthropic dario wants embedded evaluators partners up with the same company that one of his safety people left.
And let me tell you how safe the safety people are. Meter recently had a security incident. Do you know what happened?
They had a vibe coded dashboard and basically someone was able to get $600 ,000 worth of inference, by the way, donated to them and was using it. And none of them caught on because they're like, oh, we usually do usage like this. $600 ,000 worth of donated credits was used by an unauthorized user.
These are the trusted third party embedded evaluators. Now, another point from this David Sachs guy, I'm just going to play this video. I want you guys to watch it.
I got bad news for you, Chamath. an open source ban is coming. They're not going to call it that.
They're going to say that we simply have to apply the same standards to open models that we apply to close. Here's how they do it step by step. Let me explain how regulatory capture actually works.
So first of all, you got to get this regulatory apparatus. Dario wants an FDA for AI, but he doesn't have enough political support for that. So instead, they do this Trojan horse of a FINRA for AI.
They call it self -regulating. It's not really. But anyway, that gets them off the ground.
Now they've created the standard setting organization. Now they've got pre -release model testing. Then the pressure grows to codify that in law.
So that happens next. And then what they do is they say, look, all these standards need to apply equally to all models. But here's the problem with that.
Open models and closed models are technologically different. Once you release an open model into the world, you can't roll it back and you can't monitor exactly how people are using it because they run it on their own hardware. Dario says this is what makes open models dangerous.
So what they're going to do is they're going to have the standard setting body say, well, We have to set the standards for AI safety. By the way, Dario and OpenAI, they're going to fund the whole thing.
They're going to contribute all the compute. They're going to be behind it. They're going to be the ones coordinating with the government officials.
Because frankly, people in government have no idea how to monitor and control and set standards for AI safety. Technologically, this is way beyond them. So they're going to go to these companies and say, tell us how to do it.
And so what will happen is the standards will get set and then it'll be. a very simple matter of fairness to say that the standards need to apply to open as well as closed models. The open models cannot comply in the same way and gradually they will be shut out of the market.
that's basically the playbook it's as simple and as easy as that and you know what almost this feels like i know i made the mafia analogy earlier but this almost feels like the new york mobsters the new york mafia coming together and forming the commission where you have five separate families acting as one setting the rules setting the laws setting who gets hurt who doesn't it literally feels like the mafia all the stuff that's going on now there are other people who say this is a money thing they're running out of money and this is how they change stuff of this is how they avoid being profitable etc etc i don't really buy that i think based on all the evidence that i showed you that this is all about control controlling open source open way models now here's the thing i'm not saying ai is not capable of hacking and all the cyber stuff that is mentioned here but what i am saying is the people who want to be in charge of controlling who gets what and what gets where i just don't trust these people right and i think the best way to defend oneself
is by having open source models, open weight models. That's what Hugging Face used to defend themselves, right? But if only two, three private companies have access to the best intelligence in the world and everybody else has none, that just feels weird and suspect.
And I personally just don't trust the individual set before us to be the quote unquote arbiters of intelligence. At the end of the day, I have three closing remarks. First and foremost, that this is all about control.
Second of all, none of these guys are your savior. The only savior is Jesus Christ. Get right with him.
All this partnership and friendship stuff is weird to me. And last but not least, download some local models. You know, there's very smaller ones that you can run.
Download the weights. You never know. Above all, though, take advantage of all the intelligence you have access now.
Take advantage of all the subsidization you have access now. Who knows if we'll have access to these things or how much we'll have access. to and get into local models.
Download a very small model. Run it on your local machine. Maybe buy a little Linux box.
Start to sort of be self -sufficient. That is the point of this video. Now, for anyone watching that's a lawyer that's looking to sue, allegedly, right?
Allegedly. I love you guys. You've been awesome.
My name is Ross. Thank you so much for watching this video. Like, comment, subscribe.
I'll see you in the next one. Peace.
The Hook
The bait, then the rug-pull.
Three billionaires who usually can't stand each other suddenly agree on AI safety within the same week, and Ras Mic isn't buying that it's a coincidence.
Frameworks
Named ideas worth stealing.
02:42list
Dario Amodei's Pace-the-Frontier proposal
Embedded independent evaluators inside AI labs
Pacing/alignment among democracies (US + G7)
Global pacing, centered on restricting AI chip access to China
The three-part plan Dario Amodei proposed in his 'Pace the Frontier' article, which the video argues concentrates oversight inside a small circle of insiders.
15:26model
The regulatory-capture playbook (per David Sacks)
Push for an FDA-style AI regulator (fails for lack of political support)
Fall back to a self-regulating body, a FINRA-style standard setter
That body becomes the official standard-setting organization
Pre-release model testing gets built into the standard
Pressure grows to codify the standard into law
Standards are declared to apply equally to open and closed models
Open models can't structurally comply the same way closed models can
Open models are gradually shut out of the market
The step-by-step sequence the video argues is already underway, funded and coordinated by the same closed-model labs pushing for it.
Steal forRecognizing the pattern in any industry where an incumbent proposes 'safety' standards that a smaller, open competitor structurally can't meet.
CTA Breakdown
How they asked for the click.
VERBAL ASK
18:10subscribe
“Like, comment, subscribe. I'll see you in the next one.”
Standard sign-off delivered right after the substantive closing argument (download local models, trust no lab), so the ask lands after the payoff rather than before it.
Add Modern Creator as a preferred source and Google shows you more of our breakdowns in Search, Top Stories, and AI Overviews. It only changes what you see, and you can undo it in your Google settings anytime.
Add to Preferred SourcesOpens your Google source preferences with us pre-loaded. Tick the box and you're done.
A working developer walks through why two popular self-hosted agent platforms kept breaking, then rebuilds his personal agent on a file-based framework and lands on a bigger lesson about tools versus frameworks.
A creator who pays for every major AI coding subscription ranks Codex, Claude Code, Cursor, and Devin on models, subsidies, and harness quality — then reveals how to get your employer to cover the bill.
How one developer chained a computer-use verification loop and an automated code-review loop to ship 75+ pull requests with an AI cloud agent, without reading most of the code.
00:20
00:37
00:42
01:00
01:13
01:27
01:42
01:53
02:10
02:24
02:27
02:53
03:11
03:35
03:53
04:09
04:18
04:26
04:49
04:55
05:13
05:40
05:49
06:00
06:16
06:31
07:00
07:13
07:24
07:40
07:53
08:01
08:20
08:33
08:48
09:00
09:13
09:22
09:37
09:46
10:20
10:33
10:46
10:54
11:13
11:30
11:40
11:52
12:21
12:38
13:00
13:08
13:24
13:35
13:53
14:06
14:20
14:33
14:46
15:00
15:13
15:40
15:51
16:04
16:14
16:33
16:41
16:53
17:08
17:36
20:41
19:22
Video of the Day20:40
21:24
17:06
Video of the Day17:29