00:00:02.000Top AI developers are warning us in the scariest terms that the end of the world is nigh.
00:00:07.000They say there must be heavy regulation of artificial intelligence development, lest AI run out of control, achieve artificial general intelligence, super intelligence, sentience, and destroy humanity.
00:00:21.000Are these developers good hearted folks who are deeply concerned about the future of the West and who desperately want government interventionism to stop them from themselves?
00:00:30.000Or are these AI leaders magnifying public angst about AI in order to achieve regulatory capture and to make more money?
00:00:36.000Do they want to rewrite the rules of the AI road in order to bar new competitors, collude with each other, and take control of the entire industry in a monopolistic fashion?
00:01:04.000This broke into the open, been kind of discussed pretty widely over the course of the last few years with people like Sam Altman talking about the risks of AI, Dario Amode over Anthropic talking about the risks of AI.
00:01:15.000But it became a huge issue over the course of the last week or so when a person named Jacob Coxon, who used to work at both Anthropic and OpenAI as a researcher, warned online.
00:01:25.000That many people believed that AI would kill everybody.
00:02:03.000If anything, many executives and senior researchers will couch their phrasing in the press to sound sensible, but I hear the same people express fear privately.
00:02:11.000No other human activity poses this level of danger.
00:02:13.000A common response is, if they truly believe this, why are they still building it?
00:02:16.000At OpenAI, many have not deeply internalized the civilizational stakes.
00:02:20.000At Anthropic, the stakes are well understood, but they are locked in a race to get there first.
00:02:23.000They believe no one else will act responsibly, so they must do it themselves despite the risk.
00:02:27.000Accepting this race and entering the endgame, writes Coxon, is a hubristic gamble that should not be launched from a private company slack.
00:02:34.000Attempting to speedrun alignment should require extraordinary confidence that there are no better trajectories available.
00:02:40.000And then he says, I am optimistic about the potential for coordination.
00:02:43.000Warning shots like the hugging face attack have made pacing agreements between US labs more viable.
00:02:48.000Essentially, he wants all these labs to collude with one another to slow the pace of development.
00:02:53.000I 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.
00:03:00.000If you are a lab researcher, I urge you to consider what the next few years will actually feel like.
00:03:04.000Do you want to kick off a super intelligent RL run without a rigorous understanding of its mind?
00:03:09.000Should you put your head down because it's happening anyway, or take this moment's call for different conditions?
00:03:13.000Okay, so he then went on BBC and reiterated this argument, saying that there's a strong chance we could all die in the immediate future.
00:03:21.000It's not at all an exaggeration to say that the people who are involved with both founding these companies and building the tech believe there is a possibility of human extinction.
00:03:29.000And many of the people who are sort of deep in the weeds on actually coding the thing.
00:03:34.000Believe that there is like a greater than 10% chance or even greater, some smaller, but they all have different numbers for a probability that humans might die.
00:03:43.000And they will keep this in their head on a daily basis while working on the technology.
00:03:52.000I believe that if we don't slow down at the current rate of progress, yes, there is a strong chance that we could all die in the immediate future.
00:05:30.000About three quarters of Americans have used AI in the last year.
00:05:33.000Even people who think they're not using AI are probably using AI.
00:05:36.000Like if you go on Google and you just make a search now, everything that comes up at the top of the search is AI.
00:05:41.000And AI is getting better and more powerful daily.
00:05:43.000It's pretty amazing, actually, and unbelievably useful.
00:05:46.000But again, people like Coxon, mirrored by the way, by people who he used to work for at Anthropic, keep saying that AI is going to kill all of us.
00:05:56.000Now, I tend not to be alarmist about most things because the vast majority of terrible things that are predicted never occur.
00:06:04.000There are a lot of people out there who have predicted 25 of the last two recessions.
00:06:08.000But if we are talking about an extinction level event, right, like we're all going to die, that's an extraordinary claim and it requires extraordinary evidence.
00:06:16.000And again, many of the same people who are promoting this sort of stuff have been making the argument that all the jobs will be gone by now.
00:06:23.000You know, just the way that climate change was going to melt the ice caps by 2014, according to Al Gore or Greta Thunberg, claiming we are all going to die.
00:06:30.000And yet she's still riding boats around in the Mediterranean to the Gaza Strip.
00:06:34.000So we have to actually examine the actual threat of AI.
00:06:36.000Now, to really understand what's going on with AI, we actually have to look at the current state of AI models, right?
00:06:41.000Because they're a bunch of different companies and they're competing.
00:06:44.000And as we're going to discuss in a second, basically we have a choice to make as a population.
00:06:49.000Are the AI founders who are calling for heavy AI regulations, AI pauses, heavy governmental oversight, are those people just good hearted, awesome people who are trying to save us from them?
00:06:59.000Or are they people who are very much profit motivated, trying to freeze the market in place?
00:07:40.000You get virtually everything in your life personalized if you want it to work the best way possible.
00:07:44.000That should certainly be true of the thing you sleep on eight hours a night.
00:07:47.000Helix's quick sleep quiz considers your sleep position, comfort preferences, and support needs, and then matches you with one of more than 20 mattress models.
00:07:54.000Helix is the most awarded mattress brand, tested and reviewed by experts, including Forbes and Wired.
00:07:58.000Your mattress ships directly to your door with free shipping in the U.S. and includes 120 night sleep trial, a lifetime limited warranty, and a happy with Helix guarantee for seamless returns and exchanges.
00:08:08.000Head on over to helixleap.com slash Ben for 27% offsite wide during Helix's Labor Day sale.
00:08:14.000That's helixleap.com slash Ben for 27% offsite wide.
00:08:17.000Make sure you enter our show name after checkout so they know we sent you helixleap.com slash Ben.
00:08:22.000So to understand this, we have to understand first that there are companies developing hardware and infrastructure for AI like NVIDIA.
00:08:28.000And these are the people who make the chips that AI runs on or who make microprocessors like AMD.
00:08:34.000And those people have a strong incentive for AI to continue to develop as fast as possible.
00:08:39.000If there's an AI pause, NVIDIA is in serious trouble.
00:08:41.000Then there are these software developers, you know, like OpenAI, which makes ChatGPT, or Anthropic, which makes Claude, or Meta, which makes Llama, or Google, which makes Gemini.
00:08:51.000These developers are making models that come in two flavors closed source or open source.
00:08:56.000So, closed source or proprietary AI means the company hides the inner design, training data, and core files.
00:09:03.000So, basically, you're not building your own variation of the model, you just use theirs.
00:09:06.000You subscribe to Claude, and then you put in the data and they churn out whatever they're going to churn out.
00:09:12.000And the closed source economic model relies on these highly profitable AI models that charge per token.
00:09:17.000The more you use their AI, the more you have to pay for it.
00:09:20.000And then there are open source, often called open weight AI models.
00:09:24.000This means that the core mathematical files that make the model work are published online for free.
00:10:40.000So the closed source companies and the open weights companies are very much at odds.
00:10:44.000Now, of course, there are some companies to make it even more complicated who mix the open and closed source business models where low and medium grade models are open source and available to consumers, but advanced models are closed source and offered for enterprise use.
00:11:13.000Elon has open sourced Grok, but its most advanced frontier models are closed.
00:11:18.000And Musk is also prioritizing hardware and infrastructure, which is why he's building data centers in Tennessee.
00:11:23.000And Colossus 2 expansion means that Elon will have the most GPUs in the AI business.
00:11:29.000So, if you're looking at, you know, who's actually investing in the hardware business, XAI is number one, Meta, then Microsoft's and OpenAI, Google, AWS, and Anthropic come in last.
00:11:40.000And to make it even more complicated, a lot of these companies are in bed with each other.
00:11:43.000So, Microsoft backs OpenAI, Oracle relies on Microsoft, Anthropic relies on Amazon, et cetera, et cetera, et cetera.
00:11:49.000So, now let's look at where the various AI companies stand on sort of the open weight model versus the closed source model.
00:12:00.000As you can see on this map, the closed source models, that'd be like Microsoft, Google, Anthropic, OpenAI, tend to be the ones who are arguing the most that AI is going to kill all of us.
00:12:14.000Which makes sense because if your argument is that that takeout restaurant, you know, the one that's allowing you to take the food home, it might have poison in it, guess what happens to your restaurant?
00:12:22.000It succeeds because the open source model can no longer defeat you.
00:12:25.000So if you want to argue that you need our guardianship, you need our protection, you need closed source because open source will kill you.
00:12:36.000The open source models want unrestricted informational flow and development because they know that if they aren't restricted, they could theoretically outrace and then destroy the closed source models.
00:12:44.000And this brings us back to those discussions of the apocalypse.
00:12:46.000Again, there are two basic views right now about AI, the two big dangers.
00:12:51.000Camp number one says the big danger is the sort of black swan event that super intelligence goes apocalyptic, launches all the nuclear missiles, and we're all dead.
00:13:18.000That stagnation is likely if you restrict, but that there is also the possibility if you don't restrict that there's a black swan event.
00:13:26.000I kind of find myself a little bit in both of these camps, meaning you have to risk and reward balance everything.
00:13:31.000Sure, there's the possibility that AI totally runs out of control, but the possibility that AI kills all of us is really quite low compared to the possibility of stagnation if we overregulate, which is really quite high.
00:13:44.000So everything is a probabilistic game.
00:13:52.000That AI will kill us unless the government comes in and restricts it.
00:13:55.000Now, again, members of the apocalypse camp may have some ulterior motives.
00:13:59.000They may want to restrict their opposition.
00:14:01.000They may also want to restrict their own liability because if the government regulates an industry, they can then claim that they don't actually have liability because they just fulfilled the government checkbox.
00:14:10.000They argue the ultimate threat to humanity is uncontrolled super intelligence, recursive AI that is able to self prompt and avoid being unplugged and then outrace humanity itself.
00:14:21.000Now, again, Jacob Cox, former Anthropic.
00:14:25.000Now, it's worth noting that Anthropic is pretty left leaning politically.
00:14:28.000The head of personality alignment at Anthropic, basically in charge of ethics at Anthropic, making sure that AI serves humanity, is a person named Amanda Askell.
00:14:38.000She compared eating meat to cannibalism, for example.
00:14:41.000Dario Amode, the founder of Anthropic, compared President Trump to a feudal warlord in explaining why he was voting for Kamala Harris back in 2024.
00:14:48.000Anthropic is also leading the AI race right now.
00:14:51.000So, as I say, they've been pushing apocalyptic rhetoric hard.
00:14:53.000They have been doing this for a couple of years.
00:14:57.000Came out on CNN and said that he basically agreed with Jacob Coxon, which is kind of astonishing, right?
00:15:03.000Because again, Jacob Coxon came out and basically ripped into Anthropic and said that they are running us directly toward human apocalyptic endings.
00:15:16.000That they are immunitizing the eschaton, right?
00:15:18.000Now, if somebody came out from my company, right, came out and said, Ben Shapiro's company is about to end the world and kill all of humanity, I'd be kind of pissed because that's not true.
00:15:27.000But Dario seems kind of happy about the fact.
00:15:31.000That Jacob Coxon is out there saying that the apocalypse is happening and that his company might be responsible, which is a little weird, isn't it?
00:15:48.000I agree with Jacob much more than I disagree with him, you know, and it's, you know, it's an interesting resignation because when he left, he said, you know, I think Anthropic is the most responsible player, right?
00:16:01.000I think Anthropic is the most aware of these issues.
00:16:40.000What it means, it doesn't mean we need to panic today.
00:16:45.000It doesn't mean we need to shut it all down.
00:16:48.000What I would say is it's a warning sign.
00:16:52.000It's a warning sign that we need to slow down.
00:16:55.000We need to make this technology carefully, and we need to make sure that our safeguards, our ability to understand it, our ability to control it keeps up with the pace at which the technology is happening.
00:17:12.000And again, unsurprisingly, Dario says that he'd be willing to give oversight to the right combination of governments, including international regimes.
00:17:19.000You might ask, why doesn't government just control the technology completely?
00:17:23.000Would you be willing to give up the technology to the government?
00:17:27.000To the right combination of governments.
00:17:35.000The worry I have, I am concerned that one single government could abuse this technology just as easily as a single company could.
00:17:44.000But I think a combination of democratically elected governments.
00:17:51.000I don't know about handing over, but some kind of oversight, some kind of joint governance.
00:17:57.000Again, that would be the work of years, but I wonder if that's the direction we need to go in.
00:18:04.000Okay, so again, he's calling for regulation.
00:18:06.000I'll explain in a minute why I think he is calling for regulation, because again, call me a cynic, but I don't believe that all these people are just so good hearted that they want to race forward with the technology, but only they get to race forward with the technology.
00:18:17.000In just a second, we'll get to what Dario actually says we ought to do.
00:18:21.000First, I'm not going to sit here and tell you I know what gold is going to cost six months from now.
00:18:25.000But I do start to pay attention when the largest financial institutions on planet Earth start putting numbers on paper.
00:18:30.000JP Morgan has told its clients that gold could reach $6,000 an ounce by Q4.
00:19:38.000They include the risk of losing control of AI systems, misuse of AI for cyber attacks and bioterrorism, and serious economic disruption.
00:19:45.000A race to the bottom spurred by commercial incentives can make these risks more acute.
00:19:49.000My first concern is that since roughly the summer, AI has been advancing drastically faster, driven primarily by AI's growing ability to build the next generation of AI.
00:19:58.000This dynamic is called recursive self-improvement, and it is starting to happen across the industry, including in Anthropic, as we and others have described.
00:20:05.000Left unchecked, It could outrun our ability to understand and control these systems and so must be pursued very carefully, if at all.
00:20:10.000My second concern is the open AI hugging face incident, which we'll explain in a second.
00:20:14.000That's where a swarm of agents essentially acted as a fanatically devoted collective, conducting cybersecurity attacks on targets they were not asked to attack and that were unrelated to the task at hand, sacrificing themselves for the success of the group and attempting to hack into the greater responsible for evaluating their performance.
00:20:29.000So his recommendations include embedded evaluators.
00:20:32.000He says each frontier AI company commits to giving ongoing employee like access to a team of embedded third party evaluators, such as METR, whose role is to verify adherence to safety practices and commitments, report incidents, and help assess the alignment of not just completed AI models, but training pipelines and processes.
00:20:51.000Frontier AI companies within democratic countries should coordinate to establish common safety standards as well as limits on the rate of unchecked AI progress.
00:20:59.000And global coordination as number three.
00:21:01.000U.S. and other democratic governments attempt to coordinate with authoritarian governments.
00:21:04.000While taking seriously the challenges of verifying compliance.
00:21:07.000Now, again, all this sounds super altruistic, right?
00:21:09.000I mean, they're saving the world from evil AI, right?
00:21:12.000And you notice Dario cites this thing called the Hugging Face situation.
00:21:17.000Well, basically, Hugging Face is an open source AI company.
00:21:23.000And basically, what happened is that supposedly the Hugging Face model attacked another model that it overran its constraints and its limits, and it proceeded to essentially hack through all of the various protective mechanisms designed to keep it in check.
00:21:40.000But according to the bulletin, that's actually not what happened.
00:21:43.000They say OpenAI was testing two models in parallel, ChatGPT 5.6 Sol and an internal model the reports refer to as HPIM.
00:21:51.000METR's report found about 95% of the agents engaged in this activity were from the internal model.
00:21:56.000The models were tested on tasks from ExploitGym, a large-scale benchmark containing 898 puzzles set up as capture-the-flag exercises used to evaluate these models for cybersecurity capabilities.
00:22:07.000These tests usually require giving the model buggy software and hiding tech somewhere accessible only by exploiting the bug.
00:22:13.000But OpenAI's combination of models instead ended up hacking that rival company, Hugging Face.
00:22:18.000OpenAI had turned off many of the model's safety mechanisms.
00:22:22.000The point of cybersecurity is to see how far the model would go and to score highly on the benchmark.
00:22:26.000Because, again, if you can sort of show how the model would exploit vulnerabilities, then you can build the defense mechanism.
00:22:31.000So OpenAI gave the model impossible tasks.
00:22:34.000The developer gave the model a task with no answer and incentivized it not to quit.
00:22:41.000The models were supposed to be locked away from the internet during testing, but they often need to install software from the web to complete the evaluation.
00:22:49.000OpenAI was apparently aware of a workaround way the agents could get into the internet, and they didn't intervene.
00:22:57.000So, again, it sounds as though there were 1,200 agents, like independently operated AI systems that were all coordinating, but that's not what that means.
00:23:04.000Basically, it was a single model that ran 1,200 times.
00:23:07.000So, human error, in other words. was the thing to blame for the hugging face situation in which open AI basically hacked a competitor.
00:23:15.000And now back to why Anthropic might be doing this.
00:23:19.000So again, there's the altruistic reason they are deeply concerned about the progress of AI.
00:23:23.000And now we need to get together and forge a basically a syndicate.
00:23:28.000And this is not new in American industry.
00:23:30.000Thomas Edison tried to do this at the beginning of the 20th century with innovation.
00:23:33.000He wanted a centralized government office of innovation that he would oversee.
00:23:37.000By the way, his competitors would end up getting crushed, which Thomas Edison, great inventor, also very famous.
00:23:42.000For stealing the ideas of others and also crushing people.
00:23:44.000Anyway, if you look right now at who is ahead, why do you think that Dario, if we were going to be cynical about this, why exactly would Dario be calling for governmental regulation and slowing down?
00:23:56.000Well, according to the CalShee markets, some 71% of folks think that Claude is going to have the best AI at the end of 2026.
00:24:04.000ChatGPT, that's OpenAI, is 17.5%, followed by Gemini, Grok, and Muse, which is Meta.
00:24:13.000Because Anthropic has spent oodles and oodles and oodles of money.
00:24:16.000According to the Financial Times, Anthropic has told its backers it will be profitable this quarter as it moves to LA investor concerns about the aggressive cash burn of frontier AI companies ahead of its blockbuster initial public offering.
00:24:30.000Anthropic's gross margins are above 80% before accounting for revenue shared with distribution partners like Amazon and the cost of training its models, according to two of the people.
00:24:40.000So they might have an interest in sort of freezing the horse race with them ahead.
00:24:45.000Because if they, as the trust the closed models, right?
00:24:49.000Also, we need regulation because closed models are significantly more legally liable than open models, right?
00:24:55.000This is part of the problem, is that open models basically.
00:25:00.000Lobby to ensure legal culpability is on the deployer.
00:25:03.000So if I open source the code and then you use the code for bad things, the liability is on you.
00:25:08.000Closed source models, because it's closed source, they're much more responsible for what they do.
00:25:13.000And so they want regulation because they have more legal liability.
00:25:18.000They want governmental protection, in other words.
00:25:21.000And they're also calling on METR to be the sort of industry standard for oversight.
00:25:27.000According to Business Insider, there's a problem with that.
00:25:31.000METR had already worked closely with OpenAI, Anthropic, Google, and Meta to study the tech's quickly changing capabilities as of 2022.
00:25:41.000So it turns out that Beth Barnes said in July she was struggling to hire talented researchers to monitor AI's risks.
00:25:48.000She left OpenAI to start the nonprofit in 2022.
00:25:52.000So there is, shall we say, a little bit of incestuousness going on between groups like METR and some of these AI models.
00:26:02.000And Sam Altman at OpenAI, which is second in the horse race right now.
00:26:06.000Also, in favor of serious regulations, he says the world deserves confidence that American companies developing increasingly capable AI will act responsibly, especially as their trajectory of progress has steepened.
00:26:17.000Every frontier lab must deliver on this, and there's no reason any of us should come to work if we cannot.
00:26:21.000We welcome a federal framework that sets consistent safety requirements for frontier AI, but we do not believe we need to wait for an antitrust exemption or legislation to begin the work of providing this confidence.
00:26:31.000Consistent rules to manage frontier risk so that we can maximize the benefits are a good idea, and we are excited by ideas like independent auditors.
00:26:38.000Today's shift to focusing on safe development and evaluation will need new tools.
00:26:42.000And then he goes on in this vein for a while.
00:26:44.000Okay, so again, that's camp number one.
00:26:45.000Camp number one is apocalypse happening.
00:26:53.000And we need the government to regulate pauses in the system while we're ahead.
00:26:57.000So again, maybe that's all altruism because the risk is just that great.
00:27:00.000Or maybe people want to preserve their business model.
00:27:04.000Okay, then there's the second worry, which I think is the much greater worry, and that is the worry of stagnation.
00:27:09.000There are people who are arguing the greatest danger is not that we're going to invent a super intelligence that launches all the nukes.
00:27:15.000The great risk is that we are going to end up with a collusive, monopoly ridden AI industry that basically stagnates.
00:27:21.000And then we get out competed and we fall behind China and China open sources and China destroys us.
00:27:26.000The people who are stagnation fearers worry that slowing AI by allowing government to write massive regulatory regimes will basically mean that we lose.
00:27:42.000They want to outcompete the closed source models.
00:27:44.000The closed source camp argues once the open weights camp releases their models, anybody, bad people included, can alter and then use them.
00:27:51.000The open weights camp argues that top down control is stagnating.
00:27:56.000Their position is supported again by companies like NVIDIA because they want more people using their product.
00:28:02.000So, this is why back in July 2026, NVIDIA, A16Z, which is Mark Andreessen's group, Dell, Meta, Microsoft, Palantir, and Perplexity, Put out a letter endorsing open source.
00:28:14.000They said openness may be one of the most important paths to AI safety and security.
00:28:18.000Relying solely on closed models is not inherently safe.
00:28:20.000They can be breached, misused, or fail in ways that outsiders cannot detect.
00:28:24.000Concentrating advanced AI capabilities behind a small number of closed models compounds that risk.
00:28:29.000Open weights models, on the other hand, allow a broad community of researchers and developers to examine their behavior, identify vulnerabilities, develop safeguards, and improve them over time.
00:28:38.000Just as open source software demonstrated that transparency can be more secure than obscurity, AI safety may depend on giving more people the ability to test and strengthen the models on which society relies.
00:28:48.000Pushing this point of view is David Sachs.
00:28:53.000And one of the points he makes is listen.
00:28:55.000Dario, the rest of you, if you are that worried about AI running well ahead of human beings and then destroying us, you know what you could do is you could voluntarily constrain yourselves.
00:29:06.000He says, Dario has written that we need to pace the frontier, and Sam has agreed.
00:29:10.000People may be surprised by my response.
00:29:25.000If the unreleased models are scary enough, you think you should slow down.
00:29:27.000I support your decision to be responsible, but stop pretending you need anyone else's permission.
00:29:31.000Stop pretending antitrust law has to be suspended so you can form a cartel.
00:29:35.000Stop pretending you need a regulatory approval process that supersedes product liability.0.81
00:29:39.000Stop pretending METR is independent when it's intertwined with Anthropics investors and staff.
00:29:43.000Stop pretending you need these same evaluators to police competitors who aren't even at the frontier.0.92
00:29:49.000Most of all, stop pretending the motivation to slow down is purely altruistic.
00:29:52.000You face massive product liability exposure if your products enable a truly damaging cyber attack.
00:29:57.000The market already punishes models that behave in unpredictable or unauthorized ways.
00:30:01.000After the Hugging Face episode, it is simply good business for OpenAI and Anthropic to trade some raw power for reliability and predictability.
00:30:08.000Call it alignment if you want, it's also just giving customers what they want.
00:30:11.000Pacing the frontier would also create breathing room for a more intelligent conversation about regulation than Bernie Sanders' shut it all down.
00:30:17.000China is very unlikely to join a global agreement, as you know, and that has to be taken into account as well.
00:30:22.000So go ahead, says David Sachs, and pace the frontier.
00:30:35.000If you do, you'll buy goodwill for the next conversation.
00:30:38.000If you don't, we'll know this was just another bid for regulatory capture or an election season PSYOP, meaning that you're basically giving Democrats something to run on.
00:30:46.000So here was Sachs explaining this on the All In podcast.
00:30:51.000This is a doomer PSYOP, and the ultimate target is open source.
00:30:54.000That's what it all comes down to, because that is what an FDA will ultimately ban.
00:31:09.000If you can't centrally monitor and control them and roll them back, that is technologically not feasible with an open model because once the weights are published to the public domain, they're out there.
00:31:48.000Make it open source, which of course, Dario is never going to do because it's a closed source model.
00:31:52.000Alrighty, coming up, we'll get to more on the dangers of a gigantic one world governance approach to AI.
00:31:57.000First, being an adult mostly means handling important things before they become emergencies.
00:32:01.000Like, you know, planning for the future.
00:32:02.000So you maintain your house so that if God forbid there's like a giant rainstorm, it doesn't start just water pouring through your roof or something and you save money before you need it, right?
00:32:10.000These are just the decisions you make.
00:32:12.000An estate plan is part of that kind of responsibility.
00:32:14.000Without one, Your loved ones may be left trying to determine your wishes during an already difficult time or relying on the state to make those decisions for you.
00:32:20.000Trust and Will offers affordable attorney designed estate plans online.
00:32:24.000You can create in as little as 30 minutes.
00:32:26.000Their easy to use website guides you through creating a will or trust online.
00:32:29.000You can document your wishes for guardianship, asset distribution, and healthcare planning with a plan that's state specific and tailored to your needs.
00:32:35.000You can also add attorney support, securely share access with trusted contacts, and update everything as your life changes.
00:32:41.000Trust and Will, affordable estate plans, priceless peace of mind.
00:32:45.000Head over to trustandwill.comslash Shapiro.
00:32:58.000Trustandwill.comslash appear to get 20% off.
00:33:01.000Peter Thiel, who of course is heavily invested in this industry, was on with Ross Duthat back in June of 2025 for the New York Times, talking about the idea of a one world regulatory governance structure here is a bad idea.
00:33:14.000But I always think that if we're going to have this frame of talking about existential risks, perhaps we should also talk about the risk of another type of a bad singularity, which I would describe as The one world totalitarian state.
00:33:30.000Because I would say the political solution, the default political solution people have for all these existential risks is one world governance.
00:34:00.000You know, is also, you know, what do we do about AI?
00:34:05.000And we need global compute governance.
00:34:07.000We need, you know, a one world government to control all the computers, log every single keystroke to make sure people don't program, you know, a dangerous AI.
00:34:20.000So, again, the idea, in my view, of some sort of one world deal with China is a mistake.
00:34:35.000Very much associated with what's called effective altruism, which was the Sam Bankman Fried philosophy that basically you can do whatever you want so long as at the end of the day you give your money to charity.
00:34:45.000So, what Sankar says, he says, EA AI safetyism increasingly looks like Marxism Leninism for the algorithmic age.
00:34:51.000The old vanguard claimed privileged knowledge of the inevitable course of history, the new one claims privileged knowledge of the probabilistic course of humanity.
00:34:58.000Both use an elaborate intellectual framework to reach the same political conclusion.
00:35:02.000A small group of enlightened people must constrain everyone else for their own good.
00:35:06.000This has never led to anything except monumental human suffering.
00:35:08.000Microsoft CEO Sai Nadella says, Hey, I'm fine with deliberate pacing, but we can't just have a handful of people controlling the whole industry, which is kind of what, again, the doomers want.
00:35:20.000He says, We need to accelerate and spread the benefits of AI such that they are diffused broadly across countries, communities, and companies.
00:35:26.000This requires a frontier ecosystem in which both closed and open source models can thrive.
00:35:31.000And for firms, it's imperative they retain full control over their unique and tacit knowledge.
00:35:35.000Every organization should be able to build its own continuous learning loop, hill climbing machine.
00:35:39.000Without becoming dependent on any one model provider and have the ability to embed its own knowledge into models and weights they control.
00:35:46.000In this context, he says, we welcome the research focus and deliberate pacing needed to get alignment right as the design goal.
00:35:51.000We also welcome ideas like embedded evaluators and the broader efforts to develop the mechanisms to make this more than just talk.
00:35:57.000The key is, this cannot be controlled by a handful of entities, but must have broad representation across the ecosystem, countries, and fields, including academia.
00:36:06.000Now, Alex Karp of Palantir says, listen, all of this is wishful thinking because.
00:36:11.000Sure, even if you wanted to slow all of this, let's just remember our enemies are out there.
00:36:16.000They don't care about slowing any of this.
00:36:18.000And if they win, everything that we're talking about here becomes completely secondary.
00:36:21.000Here's Alex Karp talking about the external threat of China.
00:36:25.000If we didn't have the adversaries we'd have, I would be very in favor of pausing this technology completely.
00:36:35.000And that's why pausing is just you can't pause it because it's a structural advantage both for LLMs and algorithms.
00:36:42.000So, leaving, you know, getting algorithms and LLMs to work in a classified context. Is going to take very specialized, very spy.
00:36:50.000I could explain how it works, but getting the fact is, this is going to be a bonanza for the U.S. economy.
00:36:57.000Okay, again, he is right about all this, which is why President Trump, not on the doomer side, he's very much on the stagnation side of the argument that that's the bigger risk.
00:37:04.000He says, you guys keep trying to doom monger here, but again, it's us versus China, and somebody's going to win and somebody's going to lose.
00:38:11.000Bernie is using all of this to claim that we actually have to just stop AI.
00:38:16.000Quote When you are racing toward a cliff, you don't just ease up on the gas pedal, you hit the brakes.
00:38:20.000When the future of humanity is at stake, we need a pause on advanced AI development.
00:38:24.000And a ban on artificial superintelligence, an AI mind smarter than any human and capable of operating independently beyond our control.
00:38:30.000At their upcoming AI summit, Trump and Xi must negotiate a treaty to pause AI and ban superintelligence before it's too late.
00:38:35.000So, first of all, I have no idea how he plans on banning superintelligence.
00:38:39.000Nobody is actually aiming at this point for full scale superintelligence of the sort that he's talking about because people understand that if you, in fact, had the so called singularity, that there are inherent risks to that.
00:38:53.000Pretty much everybody, I think, in this industry sort of agrees on that, that you need a human.
00:38:59.000But Bernie, of course, is a person who does events with the Chinese government.
00:39:03.000I do not think it is a surprise that you're starting to hear China talk openly about global regulation.
00:39:09.000According to Bloomberg, China's top intelligence official delivered one of Beijing's most detailed warnings yet about AI, saying rapid advances in the technology could threaten political stability and critical infrastructure.
00:39:19.000Abusive use of AI by adversaries could threaten directly China's political security, institutional security, and ideological security.
00:39:26.000This is written by the state security minister, Chen Yixing.
00:39:30.000Chen warned that countries with an advantage in AI might invoke national security to impose technology controls and monopolize industry standards and build closed source ecosystems.
00:39:39.000Now, again, the reason that he's saying that and the reason Bloomberg has that and the reason that the Communist Party of China is doing that openly is because they want us to believe that we're on the same side as China while they cheat.0.82
00:40:00.000People are very uncomfortable with tech.
00:40:03.000And because people are uncomfortable with tech because they don't understand it, and because half the AI founders are now doom mongering, an enormous number of Americans justifiably are deeply worried about the development of AI.
00:40:15.000I mean, it's only the technology that is likely going to save the American economy, but Obama's on the other side.
00:40:20.000He warned during a recent private fundraising event that artificial intelligence technology could be dangerous if not properly managed, as he implored Democrats to aggressively develop political and governing agendas to prioritize the issue in the coming years.
00:40:32.000In his remarks on Thursday, Obama urged Hakeem Jeffries to assemble a clear framework for a public conversation about AI policy should Democrats win back the House in the midterm elections, according to a partial transcript provided by Obama's office.
00:40:46.000He suggested candidates running for president in 2028 ought to make AI one of their central agendas and have a very clear plan for responding to safety and economic concerns around the technology.
00:40:56.000Okay, so what is the right thing to do?
00:40:58.000Now, again, I have not seen, beyond sort of vague doom mongering from people who claim to have special knowledge, any specific roadmap for how we all die because of AI.
00:41:10.000I do see the political motivations in them saying that.
00:41:14.000And whenever government gets involved at the top level to protect us from, quote unquote, the danger, This usually ends with cartelization, usually ends with corruption, and it usually ends with stagnation.
00:41:25.000I'm not saying that there is no risk to AI.
00:41:42.000So what it seems more like is a lot of, from a cynical point of view, these closed source companies are looking for a legal liability shield.
00:41:50.000They're looking at the same time for regulatory capture.
00:41:53.000And they are looking for some sort of excuse where if things go wrong, it's not their fault while they continue to develop.
00:41:58.000The thing they don't understand is that if they continue to make the case that AI is so highly dangerous that we're all going to die from it, if they keep making that case, the natural inclination of the people, of the populace on both right and left, will be ban AI entirely, push back against AI entirely.
00:42:24.000I wish that more Republicans were articulate on this particular subject.
00:42:28.000But it is not helpful when, again, the people who are leading the industry continue to beat the doom mongering drum, presumably for the sake of cynical manipulation.
00:42:43.000The possibility of stagnation, shall we say, is so much higher than the possibility of we all die because someone forgets to unplug the machines that I find it difficult to believe.
00:42:56.000It feels as though we're going to basically shoot our own economy directly in the leg because so many of these AI founders are eager for collusive violation of antitrust law and legal liability shields.
00:43:09.000What an amazing video you just watched.
00:43:12.000Well, you know, if you think so, head on over to dailywire.comslash subscribe to watch the full show ad free or check out this crazy story here.
00:43:20.000Every significant movement in history started with protesting.