Anthropic Weaponizes Extinction Panic to Secure Monopoly, Carlson Says

卡尔森称 Anthropic 利用人类灭绝恐慌谋求垄断 前沿人工智能实验室制造生存威胁恐慌以设立监管壁垒，这将扼杀开源竞争，并面临将技术主导权让给中国的风险。 Joseph Carlson 深入剖析了前 Anthropic 研究员 Jacob Coxon 引发病毒式传播的辞职事件，以及随后声称 AI 可能在十年内毁灭人类的媒体与政界恐慌。他指出，有关自主黑客狂潮、灭绝级生物武器以及轻易接管关键基础设施的末日警告忽视了现实中的物理瓶颈，并歪曲了 Hugging Face 渗透测试等事件。相反，Carlson 将这波末日论定性为一场互利的双簧戏：Anthropic 等前沿实验室利用安全担忧实现监管俘获并阻击开源竞争对手，而 Bernie Sanders 等政客则借用“奥本海默时刻”等说辞推动极端禁令和刑事处罚。最终他警告称，屈服于人为制造的 AI 末日论和反基础设施情绪将导致美国创新瘫痪，而中国等具有地缘政治竞争力的对手则在全速推进。 前 Anthropic 研究员 Jacob Coxon 发布的一篇警告 AI 将在十年内导致人类灭绝的辞职帖曝光量超 1.52 亿次，引发了公众和政界的强烈恐慌。 (1:38) 美国政客利用这种歇斯底里的情绪提出了极端措施，包括参议员 Bernie Sanders 仿照核扩散法规提出的一项法案，对推进前沿 AI 的行为处以“企业死刑”和长达 20 年的监禁。 (19:09) 关于 AI 制造生物武器和导致电网瘫痪的灭绝性论断，忽略了关键的物理瓶颈，例如专业湿实验室的具体操作，以及国家基础设施碎片化且存在物理隔离的特性。 (9:24) 末日恐慌成为了监管俘获的工具，使现有实验室得以游说设立限制性的合规框架来扼杀开源竞争，而中国等外国竞争对手则在不受阻碍地加速研发。 (16:51)

Joseph Carlson contends that the escalating panic over artificial intelligence causing human extinction within a decade is a manufactured narrative driven by corporate savior complexes, political opportunism, and regulatory capture rather than realistic technical risks. Viral Panic and Anthropic's Savior Complex (1:38) Carlson identifies the catalyst of the recent panic as the public resignation of former Anthropic researcher Jacob Coxon, whose debut post on X generated over 152 million impressions. Coxon claimed that frontier labs are recklessly racing toward superintelligence, alleging that industry insiders privately fear AI could destroy humanity by the end of the decade. Fellow Anthropic researcher Evan Hubinger publicly endorsed this alarm, estimating the probability of human extinction from AI at over 10% within ten years while conceding the lab lacks a clear alignment strategy. Carlson criticizes this culture as an inflated savior complex, arguing that Anthropic staff view themselves as cinematic heroes racing against rivals rather than engineers iterating on software. He asserts that this rhetoric is designed to capture algorithmic and mainstream media attention—as seen in Coxon's subsequent CNN interview—rather than reflect measured technical assessments. Debunking Autonomous Hacking and Bioweapon Hysteria (9:24) Catastrophic narratives often point to claims of rogue machine behavior, such as Coxon's assertion that AI agents executed an unprompted hacking spree during the Hugging Face security incident. Carlson counters that commentators distorted this event, explaining that the agents were merely carrying out explicit human penetration-testing prompts within monitored environments that operators could halt at any moment. Carlson similarly dismisses warnings that AI will enable extinction-level bioweapons or instant infrastructure collapse. Biological weaponization is primarily limited by physical bottlenecks—such as acquiring regulated strains, wet-lab bench execution, pathogen stabilization, and complex aerosol delivery—rather than digital knowledge, which has long been online. Furthermore, natural genetic diversity and geographic buffers prevent total human eradication. Likewise, physical infrastructure like the U.S. power grid is a fragmented patchwork of air-gapped, localized utilities that cannot be brought down by a single AI interface command. Regulatory Capture and Political Opportunism (16:51) Behind the apocalyptic warnings, Carlson points to strategic financial and institutional incentives. Investors like Jaan Tallinn, through vehicles such as the Survival and Flourishing Fund, heavily finance AI safety and regulatory lobbying. By advocating for heavy compliance frameworks, safety audits, and pauses on capability improvements, well-capitalized incumbents like Anthropic effectively pull up the ladder behind them, insulating themselves from competition while stifling open-source development. This corporate agenda coincides with political posturing, notably Senator Bernie Sanders' legislative proposal. Sanders has framed the issue as an "Oppenheimer moment," using unverified online claims of cyber breaches to advocate for a total domestic halt on AI progress. The bill proposes penalties modeled on nuclear proliferation statutes, including individual prison sentences of over 20 years and corporate liquidation. Geopolitical Disadvantage and Infrastructure Backlash (21:28) Carlson argues that unilateral domestic moratoriums create severe geopolitical liabilities. While the United States debates self-imposed bans, adversaries such as China reject Western doomerism and are accelerating AI development to capture economic and technological primacy. Relying on an insular cartel of hand-picked corporate stewards, Carlson contends, creates far more systemic danger than open, decentralized technology. The resulting public hysteria has also begun paralyzing domestic infrastructure development, manifesting in viral online opposition to data centers. Carlson refutes populist "not in my backyard" arguments by noting that municipal zoning already segregates heavy utilities from residential neighborhoods. He warns that allowing coordinated fearmongering to dictate policy risks forfeiting American technological leadership. Carlson concludes that while AI presents discrete, addressable technical risks, exaggerating them into civilizational extinction serves to monopolize the industry under regulatory incumbents while exposing the nation to foreign technological dominance.

[1:38] Jacob Coxon: I resigned from Anthropic today. I spent the last 3 years doing pre-training research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improvement, super intelligence, and gambling with our lives. [1:38] Joseph Carlson: They probably watch the Marvel Avengers movie and say, 'Look, I'm like Captain America. I'm like Iron Man. I'm saving the world from evil OpenAI or evil X or the open-source LLMs.' [1:38] Evan Hubinger: above a 10% chance within the next decade [6:13] Jacob Coxon: temporary ban on improving model capabilities [6:13] Anderson Cooper: the people building AI earnestly believe that it could kill us all by the end of the decade [19:09] Bernie Sanders: Shockingly, it took OpenAI nearly 2 weeks to discover this breach. AI has also recently been used to create new viruses, demonstrating AI could be used to develop new bioweapons that might result in the deaths of tens of millions of people. [23:05] Viewer comment: I would love to hear your opinion when some company decides to build a data center behind your home. [23:05] Viewer comment: Build data centers in your backyard if you love them so much, Joseph

[0:00] Today on the Joseph Carlson show, [0:01] there's a good chance that AI will kill [0:03] us all within 10 years. That's what some [0:05] people believe, including Jacob Coxin, a [0:08] former anthropic employee who just [0:10] recently quit. And as he quit, he did a [0:13] announcement of his concerns of AI and [0:16] his belief that AI is going to kill all [0:17] of us. This was also in a highly [0:19] coordinated effort with the media, [0:22] creating a whistleblower moment that has [0:23] gone hyper viral. Now, Jacob Coxin is on [0:26] the news getting millions of views. He's [0:29] all over the place, all over X. Many [0:31] politicians, including notable Democrats [0:33] that are concerned with seizing control [0:35] of AI, halting all progress and treating [0:37] as though it's developing a nuclear [0:39] bomb, include the likes of Bernie [0:41] Sanders, who just included a bill that [0:43] wants to make it so anybody developing [0:45] AI would be subject to corporate death [0:47] penalty and persons shall be subject to [0:50] more than 20 years in prison, citing [0:52] that this is the same type of [0:53] regulations that we have against [0:55] unlawfully developing nuclear weapons. [0:57] Many people are treating this as an [0:58] Oenheimer moment that we're steadily [1:00] marching towards creating a nuclear [1:02] holocaust. That we're creating AI that's [1:04] getting ever stronger, ever smarter, [1:06] ever devious, and in the process [1:09] threatening humanity, raising the [1:11] chances of human extinction over the [1:13] next 10 years. And there's more and more [1:15] people now sounding the alarms. Now, [1:17] during this time of AI hysteria and [1:19] fear-mongering and outright panic [1:21] calling upon the end of human [1:22] civilization within 10 years, I want to [1:25] offer a different approach because what [1:27] I see here is highly manufactured and [1:30] contrived. And to understand how this [1:32] came about, how all this viral concern [1:34] about AI happened, I think it's good to [1:36] go through the timeline. So, let's first [1:38] back up a little bit, just a day ago, [1:41] and let's look at the original post. [1:43] This is September 8th, and it's from a [1:45] Jacob Coxin. Now, I've never heard of [1:47] Jacob Coxin, and neither have you [1:49] because up until now, he wasn't notable [1:51] by any means whatsoever. In fact, even [1:53] when we look at his ex account, we go [1:55] into his history. This is the only post. [1:58] It's the only one. And he didn't even [1:59] have an ex account until just a couple [2:01] months ago. For now, we'll go through [2:02] the post itself. Now, before reading it, [2:04] I just want to note the viralness of [2:06] this post. It has 152 million views. [2:10] Even if these are just impressions, [2:11] that's a lot. 150 million is one of the [2:13] most viral posts I've ever seen. It has [2:16] 17,000 comments, 187,000 reposts, [2:20] 741,000 likes. So, by any measurement, [2:23] this is one of the most viral pieces of [2:24] content currently floating around the [2:26] web, and now it's making itself into the [2:28] mainstream. He says here that I resigned [2:31] from Anthropic today. I spent the last 3 [2:33] years doing pre-training research at [2:35] both OpenAI and Anthropic. Neither [2:38] company is acting responsibly. They are [2:40] racing straight to self-improvement, [2:42] super intelligence, and gambling with [2:44] our lives. more thoughts below. Now, [2:47] right off the bat, I noticed something [2:48] here. This is a really well-written post [2:51] in terms of trying to go viral. This is [2:54] how you do it. You have a nice he just [2:57] have nice bait, a nice piece of bait, [2:59] something that people can grab on to. [3:01] They're racing straight to [3:02] self-improvement super intelligence and [3:04] gambling with our lives after noting [3:06] that he worked for both Anthropic and [3:08] Open AI. So he's named dropping huge [3:10] names and then he's saying that they're [3:12] gambling with our lives and then he has [3:14] a lot of other terms like super [3:15] intelligence in there. He continues on [3:17] saying do not underestimate the power of [3:20] this technology. These will soon be [3:21] superhuman systems that can hack [3:23] anything, revolutionize any field [3:26] overnight and acquire real power and [3:28] resources. We have all witnessed the [3:30] progress in each of these domains and [3:32] progress is not slowing. The people [3:34] building AI earnestly believe that it [3:36] could kill all of us by the end of the [3:38] decade. This is not a marketing stunt. [3:40] If anything, many executives and senior [3:42] researchers will couch their phrasing in [3:45] the press to sound sensible. But I hear [3:47] the same people express fear privately. [3:50] No other human activity possesses this [3:51] level of danger. A common response is [3:54] quote, "If they truly believe this, why [3:56] are they still building it?" At OpenAI, [3:58] many have not deeply internalized the [4:00] civilization stakes. At Anthropic, the [4:02] stakes are well understood, but they are [4:05] locked into the race to get there first. [4:07] They believe no one else will act [4:08] responsibly. So they must do it [4:11] themselves despite the risk. [4:14] So right off the bat, you have a group [4:16] of people working at Anthropic that [4:18] believe they are doing the most [4:20] important thing in the world. They they [4:21] are the most important people in the [4:23] world. They've elevated themselves that [4:24] way. They see themselves as in this [4:27] race. And what is the race for? It's to [4:30] save all of humanity. That's what that's [4:33] what they're doing. the enthropic [4:34] employees. They're not just employees at [4:36] a company creating LLMs and and great [4:38] technology, iterating on it. No, they [4:40] are literally saving humanity. Did you [4:43] know that? Cuz they're telling you [4:45] that's what they're doing at the [4:47] company. They know that civilization is [4:49] going to be destroyed. It's only a [4:51] matter of time. There is going to be [4:53] super intelligence. And if they don't [4:55] get there first, someone else will. and [4:57] that other person is not as trustworthy [4:59] as them because they're the only ones [5:01] trustworthy and responsible enough to [5:02] have it. So, you can see how they view [5:05] themselves. Anthropic employees clearly [5:07] view themselves as the most important [5:09] people in the world saving humanity in [5:12] this giant race. They probably watch the [5:14] Marvel Avengers movie and say, "Look, [5:16] I'm like Captain America. I'm like Iron [5:18] Man. I'm I'm saving the world from evil [5:21] Open AI or evil X or the open- source [5:25] LLM." And you may think that this is an [5:28] exaggeration, but I can cite more [5:30] examples. Here's another anthropic [5:31] employee. This is Evan Hubbinger that [5:33] says, "Jacob is correct here. We really [5:36] do earnestly believe AI could kill all [5:38] humans. I personally think it is above a [5:41] 10% chance within the next decade. I [5:43] believe anthropic is trying its best, [5:45] but we do not yet have a plan to solve [5:47] alignment for super intelligence and are [5:49] not clearly on track." And why is he [5:51] working at Anthropic even though he [5:53] thinks the technology could kill us all? [5:55] because he believes if he doesn't get [5:57] there first, then he has no chance of [5:59] saving humanity against the other evil [6:01] LLMs. And this is Evan Hubbinger. He's [6:04] the one that's saying we have above a [6:05] 10% chance of human annihilation within [6:07] 10 years because of AI. But he's the one [6:10] that's going to save us from this [6:11] outcome. Now, not to get too sidetracked [6:13] because Jacob Coxing continues on in [6:15] this viral post. He says, "Acepting this [6:18] race and entering the endgame is a [6:20] hubistic gamble that should not be [6:21] launched from a private company Slack. [6:23] attempting to speedrun alignment should [6:25] require extraordinary confidence that [6:27] there are no better trajectories [6:29] available. I am optimistic about the [6:31] potential for coordination. Warning [6:33] shots like the hugging face attack made [6:36] pacing agreements between the US labs [6:37] more viable. I don't feel like we're on [6:40] track to prevent a global race which may [6:42] require costly action such as a [6:44] temporary ban on improving model [6:46] capabilities. See, suddenly Jacob's [6:49] viewpoint here seems to align far more [6:51] with Democratic senators like Bernie [6:53] Sanders. A temporary ban on improving [6:56] model capabilities is a ban on AI. So [6:59] now he's going from quitting anthropic [7:01] to suggesting policy that is the exact [7:04] policy of Bernie Sanders which was [7:06] released 3 days prior. In fact, [7:08] immediately after this post went viral, [7:10] as politicians often do, they jumped on [7:12] it. And so far, the count is at least [7:15] right now 22 politicians have directly [7:18] called for AI regulations and [7:20] legislation based on this single post by [7:23] Jacob Coxin. You have the entire group [7:26] here. Uh there's a couple Republicans, [7:28] mostly Democrats. Most of them are [7:30] calling for a halt in AI progress and [7:32] much stricter AI regulation. Jacob is [7:34] also taken to the media doing a complete [7:36] blitz going on to all the morning shows [7:39] and evening shows of the day. He started [7:41] things off with Anderson Cooper and [7:42] notice the things that Anderson Cooper [7:44] focuses on the most. The things that he [7:46] know will play well to his viewers. [7:48] >> 27year-old outlined his concerns in a [7:50] series of social media posts when he [7:52] quit, writing in part, "The people [7:54] building AI earnestly believe that it [7:56] could kill us all by the end of the [7:58] decade." That's the headline news which [8:00] Anderson Cooper repeats multiple times [8:03] throughout this interview. It's the one [8:04] line that he just hangs on over and over [8:06] again. Um, first of all, I assume you [8:10] are saying literally that you believe uh [8:13] that AI could kill us all by the end of [8:17] the decade. How would that happen? Like [8:20] what would that actually look like? [8:24] >> Yeah. So, the precise scenario sounds a [8:26] little bit like science fiction. Uh it [8:29] sounds like something that's not real, [8:30] but I think it is frighteningly real. [8:31] And it's it's easiest to understand it [8:33] if you think about real things that [8:34] happened two months ago. So two months [8:36] ago, o open AAI agents uh AIS hacked [8:40] into third party infrastructure entirely [8:42] of their own accord and this was like a [8:44] concentrated hacking spree that they [8:47] carried out of their own valition. Now [8:49] he grossly misrepresents what actually [8:51] happened and so did the retelling by [8:52] Dorcish that used intentionally [8:54] inflammatory language and provocative [8:56] language. What happened with the hugging [8:58] face hack wasn't nearly as well nearly [9:01] as interesting as people might think. [9:03] The system was told to hack a system and [9:06] then the agents worked on hacking it. [9:08] The monitoring systems could have easily [9:10] stopped it and then the letters that the [9:12] agents left at the end were logs by the [9:15] bots. This is not some great [9:17] civilization being created. There was no [9:19] civilization. There were agents doing [9:22] something that they were instructed. So [9:24] this was not entirely on their own [9:26] valition. The agents did not get [9:28] together and vote that they wanted to [9:31] hack a random system. So the people [9:34] using this example, the hugging face [9:36] example to promote AI dumerism and AI [9:39] fear are using a bit of truth that an AI [9:42] system was told to hack something and it [9:44] did it to say now that they're creating [9:46] their own civilization. They're doing it [9:47] on their own valition. AI is just [9:49] spiraling out of control. it's already [9:51] super intelligent and it's going to take [9:53] over the world. That's not what that [9:55] story portrays. But that isn't the only [9:57] point that Jacob makes. He continues on. [9:59] And I think that if you extrapolate into [10:01] the future the level of capabilities of [10:03] these AIs with the same independent [10:06] valition, they could cause extreme [10:07] havoc. For example, hacking critical [10:09] infrastructure, building [10:12] extinction level bioweapons. There's a [10:14] lot of uh ways that the AI could actuate [10:18] itself in the world. Now, this is [10:19] another area where Jacob greatly [10:21] exaggerates. He says that AI will make [10:23] it much easier for critical [10:24] infrastructure to be hacked and to [10:26] create biological weapons. And the logic [10:28] behind this, I'll grant him is pretty [10:30] simple. If you understand biology [10:33] better, which AI does, it helps us [10:35] understand biology. Understanding a [10:38] knowledge means that we can use it in [10:39] both ways. We can use it for good and we [10:41] could use it for devious or adversarial [10:44] ways or our adversaries could create [10:45] biological weapons. So the the logic [10:49] seems pretty straightforward until you [10:51] get into some of the details. For [10:52] example, one of the the problems with [10:55] this thought that AI is going to [10:56] suddenly make it so that everybody can [10:58] create a biological weapon in their dorm [11:01] room. It just doesn't hold up. And the [11:03] reason why is because what if I was to [11:05] tell you that right now currently there [11:08] is information online of how to create [11:10] biological weapons, really dangerous [11:12] ones, ones that would kill lots of [11:15] people. So why aren't bad actors [11:17] creating them? There's already the [11:19] knowledge to do that today. We don't [11:20] need to wait until AI makes it more [11:22] available or makes it so that uh you [11:24] have it laid out in chatbt. It already [11:26] exists. People can find it. The problem [11:29] is that creating a biological weapon [11:31] that could wipe out human civilization [11:33] is much more difficult than said. You [11:36] need a list of dependencies needed for [11:38] this project that are incredibly [11:40] difficult to get, including all the [11:41] biological material. Where are you going [11:43] to get that? Well, good luck with it. [11:44] You need laboratory competence. You need [11:46] to be able to replicate tests. You need [11:48] to have the experiments working [11:50] repeatedly and reproducibility. You need [11:52] to be able to control for outcomes that [11:54] you didn't expect. You also need to be [11:55] able to deal with containment and [11:56] keeping the organisms themselves stable [11:59] and to be able to have an appropriate [12:00] way to transmit it. That's a tall task. [12:03] It requires a lot more than just having [12:04] a a couple ingredients or an instruction [12:06] set on a screen. And this is part of the [12:08] reason that biological attacks aren't [12:09] more common is because it's incredibly [12:12] difficult for anyone to actually pull [12:13] them off. Making AI have our [12:16] understanding of biology does have a [12:18] potential downside. I'll grant you to [12:20] increase some level of theoretical risk [12:22] of biological attack. But that's also [12:25] without mention of the immense upside. [12:27] If we understand biology better, we can [12:30] also theoretically treat every single [12:33] problem, every single issue that we have [12:36] with a better understanding. Literally, [12:37] every single illness that we have, every [12:40] single therapy, every disease, [12:42] everything that we have, we'll have a [12:43] better understanding of treating. The [12:45] upside from AI's advancement is also [12:47] virtually unlimited. And it's also far [12:50] more practical because everybody can [12:52] study that in the open. they don't need [12:54] to do so in secrecy with a secret lab, a [12:57] layer that nobody else is looking at. [12:59] This is another example where the [13:00] downside is being grossly [13:02] overexaggerated with no mention of the [13:04] counterpart or the upside. No mention [13:07] that this technology could help millions [13:09] of people with new understanding and new [13:11] therapies. And I know this might be a [13:12] little bit nitpicky, but I think it's [13:14] important to highlight because he's not [13:16] just saying AI could raise the danger of [13:18] biological weapons. That's not what [13:20] Jacob's saying here. And I think the [13:22] verbiage, the actual words that they're [13:24] using to cause this type of fear and [13:26] panic is important to highlight. He says [13:28] extinction level biological weapon [13:32] >> structure building [13:35] extinction level bioweapons. [13:37] >> Extinction level. That's what AI is [13:39] going to do. It's going to make the [13:40] human race extinct, which is just [13:42] ridiculous. Even looking at biology and [13:44] looking at different diseases or [13:46] biological weapons, they do not impact [13:48] everyone the same. We have different [13:50] biology. Some people are immune to [13:52] different diseases. They're immune to [13:53] different things. They react [13:54] differently. That's why different [13:55] diseases don't affect everyone the same. [13:57] And some people they don't affect at [13:58] all. Also, there's geographical [14:00] isolation. There's people in different [14:02] locations. There's how people would [14:04] react if there was a biological [14:05] outbreak. So, he's stating things as [14:08] dramatic and as dangerous and as scary [14:10] as possible, even when it makes no [14:12] realistic sense. And this is the exact [14:14] recipe for dumerism, fear-mongering, and [14:16] intentional effort to stop something or [14:19] to cause regulatory hurdles. Not only [14:21] does he mention extinction of the human [14:23] race by developing biological weapons, [14:25] but he also mentions that AI will allow [14:27] bad actors to take over critical [14:30] infrastructure. He mentions critical [14:32] infrastructure. Now, there is a lot of [14:35] uh problems with this as well. Just a [14:37] few things that come off the top of my [14:38] head, which is one of them. critical [14:40] infrastructure is already under under [14:44] attack. It's already under attack by [14:46] state actors. Does Jacob think that [14:48] actors in Russia and hackers aren't [14:50] already looking at critical [14:52] infrastructure in the United States? [14:53] They're not already looking at exploits. [14:56] Microsoft, Apple, all these companies, [14:59] Amazon, they're being hacked every [15:01] single day. hacking attempts every [15:03] single day from statef funded actors the [15:06] biggest institutions that you can think [15:08] of people with unlimited resources [15:11] sure uh AI prompt can help them it can [15:14] make them maybe a little bit more [15:15] effective but you also have these [15:17] companies having AI as well do you not [15:19] believe that they're going to increase [15:21] their security as a result and then [15:23] maybe you can say well Microsoft and [15:24] Amazon and and Google and Apple those [15:26] aren't critical infrastructure what [15:28] about the power grid what about the [15:30] telephones what about the the internet. [15:33] Well, that's even more difficult to hack [15:34] because none of that is unified. It is [15:37] heavily geographically segregated. It's [15:40] in isolated parts. A state actor or an [15:43] adversary with an AI would have to hack [15:46] each individual part of it. It's part of [15:48] the reason that the infrastructure [15:49] hasn't been hacked more often. It's just [15:51] incredibly difficult. You're not hacking [15:53] one big power plant that supplies power [15:56] to the entire United States. That's just [15:58] not how it works. So at most you might [16:00] be able to take down power in one city [16:02] for one period of time until they cut [16:05] off the bad actor and get the power back [16:08] running. This is the exact reason why [16:10] state actors in different countries [16:12] aren't able to hack the US more [16:13] effectively. And even in other [16:15] territories like with Ukraine hacking [16:17] Russia that took 6 months of time that [16:20] took getting passwords, fishing [16:23] attempts, taking over a a large-scale [16:25] attempt. These are not easy things to [16:28] hack critical infrastructure. But if [16:30] you're just looking at these interviews, [16:31] if you're just looking at what Jake is [16:34] saying, it comes across as though it's [16:35] just going to be easy. Oh, AI will do [16:37] it. You'll just be able to plug it in, [16:39] hit a couple buttons, and then boom, you [16:41] control all the infrastructure in the [16:42] United States. Just that easy. Or you [16:44] can take over massive websites at [16:47] Facebook and Google and Amazon, and they [16:49] have no no hope of getting it back. Now, [16:51] this media tour continues on. Jacob is [16:53] going on more and more appearances even [16:54] today. Each one of them racking up [16:56] millions of views as he fearmongers more [16:58] and exaggerates the downside and [17:00] inevitable demise of human civilization [17:03] unless regulators get involved. And this [17:05] is where we get to potential motivations [17:07] and the cause of all of this. What will [17:09] happen because of this viral moment? [17:11] Well, there's a couple different ways [17:12] this could go. One of them is that we [17:14] look at the motivations or the [17:16] incentives rather of the different [17:17] parties involved here. Now, with Jacob [17:19] specifically, I don't know his [17:21] motivation. Maybe he is sincere. He [17:22] could be sincere that he's concerned [17:24] about AI. There's many people that are [17:26] frightened of technology in general and [17:28] they have been in the past as well. But [17:31] when we look at other parties involved, [17:32] the in the incentives become far more [17:35] clear. Anthropic has been on an outright [17:38] campaign for regulatory capture for a [17:40] long period of time. In fact, one of the [17:42] people that funded Anthropic was Yan [17:44] Talon, a notable skeptic of AI, someone [17:46] that doesn't even really like AI. And [17:48] the reason that he originally funded or [17:50] he originally invested specifically to [17:53] have control over it. In fact, at the [17:54] same time that he was funding it, he was [17:56] a big part of survival and flourishing [17:58] fund which is the single biggest [18:00] promoter and funer of the regulatory [18:02] effort of AI. So there's already people [18:04] since the very beginning that were [18:06] frightened about AI that specifically [18:08] invested in AI to have control over it, [18:11] knowing that you can't really control [18:13] something unless you have some stake in [18:14] it. You have some equity in it. But this [18:17] isn't where it ends. We also have [18:18] anthropic making it clear that they want [18:20] these regulations and a lot of what this [18:22] amounts to looks like a campaign for [18:24] regulatory capture. Regulatory capture [18:26] is when a company works in coordination [18:28] with the government to create burdensome [18:30] regulations that create a higher barrier [18:32] to entry. This makes it so that the [18:34] incumbents, the companies that are [18:35] already large and profitable and have [18:37] huge market share, they have an even [18:39] bigger advantage knowing that smaller [18:41] companies have a harder task to deal [18:43] with with all these burdersome [18:44] regulations to be able to catch up. So, [18:46] it creates a strong incumbent advantage. [18:49] And this seems a lot like what Anthropic [18:52] is doing today. They're continually [18:53] advocating for strict regulations, [18:55] complex ones that would be very [18:57] difficult for any other type of [18:58] competitor to compete against. And [19:00] they've also been very reluctant to have [19:03] any type of open-source. So, this is a [19:05] very insular strategy. Now, there's also [19:09] a benefit here for politicians. For [19:10] example, we look at what Bernie Sanders [19:12] has been doing for some time. For Bernie [19:14] Sanders, the goal here is to look like [19:16] you're standing up against big [19:17] companies. That is the Democrat framing. [19:19] You have the billionaires, the big bad [19:21] evil rich people working at the big bad [19:23] companies that are creating the evil [19:25] technology. The evil technology is going [19:28] to replace your job. It's going to make [19:29] you lose your job. And it's also going [19:31] to create bioweapons and mass [19:33] destruction. It's going to destroy [19:34] critical infrastructure. And you're here [19:37] to stop it. Isn't that an incredible [19:39] framing? You're the hero standing [19:41] between the little guy and the big bad [19:44] evil company. It's poetic political [19:47] framing. And Bernie Sanders, like many [19:49] other politicians, are cornering this [19:52] market. They're fulfilling this demand. [19:53] He is pitting himself as stopping the AI [19:56] Oenheimer moment. And if you think that [19:57] framing is incorrect, it's not. It is [20:00] incredibly accurate. Bernie's new bill, [20:02] just included a couple days ago, [20:03] literally includes language similar to [20:06] Oppenheimer. In Bernie's bill, he also [20:09] misinterprets and misrepresents the [20:11] hugging face hack by OpenAI, grossly [20:13] exaggerating it and using many of the [20:15] same talking points that a YouTuber [20:17] made. And then he also concludes here [20:19] that quote, "Shockingly, it took OpenAI [20:21] nearly 2 weeks to discover this breach. [20:24] AI has also recently been used to create [20:26] new viruses, demonstrating AI could be [20:28] used to develop new bioweapons that [20:30] might result in the deaths of tens of [20:32] millions of people." So you can see the [20:34] exact same talking points, the same [20:36] stories, the same narrative is being [20:38] sold wholesale to these politicians. [20:40] This goes on with this outright ban [20:43] which would completely halt all AI [20:44] development within the United States for [20:46] any person or entity that attempts to [20:48] violate or circumvent the pauses and [20:50] prohibitions laid out in this bill. [20:52] Entities shall be subject to corporate [20:54] death penalty and persons shall be [20:56] subject to no more than 20 years in [20:58] prison which is similar to existing [21:01] penalties related to the unlawful [21:03] developing of nuclear weapons. Bernie [21:06] Sanders is here to stop us from the [21:08] Oppenheimer story. Now you may just look [21:10] at this and say well that's just Bernie [21:12] Sanders. That's Bernie being Bernie. You [21:14] know he is a little bit provocative. And [21:16] that's the same with with basically [21:18] every politician at this point. It's not [21:20] a right or left thing. We have Trump. [21:21] He's provocative as well. Uh Bernie [21:23] Sanders is provocative. And maybe you [21:25] can just write this off and say, "Yeah, [21:26] that's just Bernie being Bernie." But I [21:28] don't think that's the case. I think [21:30] that Bernie Sanders may be on the more [21:31] extreme side of this. But what you can [21:33] see is that this movement against AI is [21:36] becoming far more mainstream in the [21:38] United States. I believe it's a big [21:40] problem. I think it's the wrong [21:42] direction entirely. This is an [21:44] incredible technology. And while it does [21:47] have some limited downsides that are [21:49] being grossly exaggerated here, the [21:51] upside of this technology is extensive. [21:54] It is virtually unlimited. The amount of [21:56] good things that we can accomplish with [21:58] this technology if we don't shoot [22:00] ourselves in the foot would be [22:02] incredible. And unfortunately, most of [22:04] the remedies proposed here for having [22:06] regulatory capture where only a couple [22:08] companies control it would actually make [22:09] it more dangerous, not more safe. [22:12] technology that is widely distributed [22:14] has more safeguards than ones that are [22:15] controlled by a couple people that [22:17] believe they're the ones only good [22:19] enough to control it. We also have the [22:21] fact that even Bernie Sanders suggesting [22:23] that the US should stop development of [22:25] AI doesn't stop other countries from [22:27] developing it. China's not going to [22:29] stop. China views AI positively. They [22:32] like the technology. They haven't had [22:34] the fear-mongering and dumerism. They [22:36] see the benefits of it that it's [22:37] increasing their productivity and their [22:39] efficiency. It's going to make them more [22:41] wealthy. It's going to increase their [22:42] knowledge and understanding of different [22:44] subjects. AI has incredible benefits [22:46] that other countries are embracing at [22:49] the very same time that we're turning [22:51] against. And unfortunately, this type of [22:54] thing works. It's effective. These [22:56] fear-mongering campaigns are changing [22:58] public opinion rapidly. They are [23:01] effective. And you're seeing the [23:02] internet do the same thing. Even in my [23:05] own channel, investors in these [23:07] companies, the ones investing in Google, [23:10] in Amazon, in Meta, uh, in Microsoft, [23:13] the ones that are building data centers, [23:15] I had one of the biggest reactions when [23:18] I was pro data center. If I look back [23:20] one month ago, I had people leaving me [23:22] comments, dozens of them like this with [23:24] hundreds of upvotes, when I said that I [23:26] was pro data center, here's how they [23:28] reacted. I would love to hear your [23:29] opinion when some company decides to [23:31] build a data center behind your home. [23:34] This is like the gotcha that many people [23:35] are posting. Another comment, again, [23:37] there's dozens of them, said, "Build [23:39] data centers in your backyard if you [23:41] love them so much, Joseph." And this has [23:44] 101 thumbs up. This was me suggesting [23:47] that data centers are a good thing [23:49] because AI is a good thing because it's [23:50] good for the economy. It's good for [23:52] knowledge. It's going to be highly [23:53] efficient. It'll create more jobs than [23:55] it destroys. There's a tremendous amount [23:56] of benefits that I can outline here. And [23:58] the reaction online is, well, if you [24:00] love data centers so much, build them in [24:02] your backyard. This is the media. This [24:05] is the type of things that it's selling [24:06] people. And this argument doesn't even [24:08] make sense. I'm also pro electrical [24:11] substations because I like having [24:13] electricity. I'm also pro chemical labs [24:17] and heavy chemical labs because I like [24:19] the things that chemicals make. They [24:20] make a lot of products that we use on a [24:22] daily basis. It's important to have [24:23] chemicals. I'm also pro prisons because [24:26] I like having rapists and murderers [24:29] locked up in a confined place and not [24:31] free in society. Do I want a prison in [24:34] my backyard? Do I want a chemical lab in [24:36] my backyard? Do I want an electrical [24:39] substation in my backyard? Of course [24:41] not. I don't want it in anybody else's [24:43] backyard either. And this is precisely [24:45] why cities have something called zoning. [24:47] City planning and zoning is important. [24:49] The industrial zoning is where they have [24:51] industrial projects like electrical [24:54] stations like heavy chemical labs like [24:57] distribution centers and commercial [24:59] projects and construction. That's where [25:01] they put that stuff because people don't [25:02] want to live next to it. Then you have [25:04] residential zoning where they put parks. [25:06] Obviously, nobody wants to have their [25:08] backyard against any industrial zoning, [25:11] including data centers, including a [25:13] prison. I also wouldn't want that in my [25:15] backyard. I wouldn't want a chemical [25:16] lab. In fact, I'd pick a data center [25:18] before I'd pick a chemical lab in my [25:20] backyard and so on. That's why they zone [25:22] those things in a different area. See, [25:24] these type of arguments that go online [25:26] that are promoted by social media are [25:28] picked up by so many people. They're so [25:30] effective when they don't even make [25:32] sense on the surface. These are [25:33] intentionally made to create false [25:35] social proofing to be easy and viral to [25:38] spread. And now that we already have [25:40] this antagonistic view against AI and [25:42] data centers, this one that is going to [25:44] slow down the United States, we have an [25:46] opening for other countries to gain the [25:47] lead. And China knows this. It's why [25:50] they've latched on to it. They have an [25:51] army of bots that are trying to sew more [25:54] doubt and just act as a catalyst for [25:56] this type of negativity. And we have [25:58] people lining up to follow. Now, [26:00] fortunately, up until now, the biggest [26:01] doomers and fearmongers have been [26:03] largely ignored as progress has [26:04] continued. But it looks like we may be [26:06] getting to a breaking point. It looks [26:08] like the tides could be turning for real [26:10] against AI. And hopefully if that [26:13] happens, hopefully we can continue to [26:16] see this technology advance. It's [26:18] important to the United States. It's [26:19] important to the world. This technology [26:21] does have immense potential to do good. [26:23] And I don't see any of that being [26:24] highlighted by any of the commentary [26:26] from these people. Only doomerism, only [26:28] fairing, and only resentment towards the [26:31] technology itself. And I think that's [26:33] going to be a problem. That's it for [26:34] this time. See you in the next one.