Who Will Save Us From AI?

Sources

Hao, Karen. Empire of AI. https://bookshop.org/p/books/empire-of-ai-dreams-and-nightmares-in-sam-altman-s-openai-karen-hao/de10c251433f34d2

Amodei, Dario. We Must Pace the Frontier. September 2026. https://darioamodei.com/post/we-must-pace-the-frontier

Gecker, Jocelyn. Microsoft commits to sweeping AI privacy rules for students. Will other tech giants follow? AP. September 15, 2026. https://apnews.com/article/ai-school-student-data-privacy-microsoft-google-89b040631d635e0d8673f1e9aa9a6e28

Boak, Josh. Trump calls AI risks a ‘hoax,’ says there is a ‘SICK conspiracy’ against AI and data centers. AP. September 14, 2026. https://apnews.com/article/trump-ai-guardrails-data-centers-b85df16775ff7e9611a456b061a0e4b9

Marquez, Alexandra. Anthropic whistleblower says Congress should let AI companies police themselves until a regulatory body is established. NBC News. September 13, 2026. https://www.nbcnews.com/tech/security/anthropic-whistleblower-congress-ai-companies-regulation-rcna597465

Drenon, Brandon. AI regulation faces political deadlock as calls grow for Congress to act. BBC. September 15, 2026. https://www.bbc.com/news/articles/ck20989806e9o

Powel, James. House Speaker Johnson says Congress shouldn't lead AI safety efforts. USA Today. September 13, 2026. https://eu.usatoday.com/story/news/politics/2026/09/13/mike-johnson-ai-safety-developers-responsibility/91750965007/

Goode, Lauren. One of AI’s Fiercest Critics Says All the Doom Talk Is ‘Meant to Distract Us’. Wired. September 11, 2026. https://www.wired.com/story/one-of-ais-fiercest-critics-says-all-the-doom-talk-is-meant-to-distract-us/

Hardy, Annie. Post on LinkedIn. September 15, 2026. https://www.linkedin.com/posts/anniehardy_my-comments-on-amodeis-letter-slowing-down-activity-7505260108084514817-8KwU/

Greenblatt, Ryan et al. Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident. METR. August 26, 2026. https://metr.org/blog/2026-08-26-openai-hugging-face-incident-investigation/#core-takeaways-about-this-incident

Transcript

Thank you to Incogni for partnering with me on today’s video! More on them later.

In July, OpenAI researchers were running tests, using tens of thousands of AI agents each operating separately in a sandbox environment. They were blocked from accessing the internet and they did not know about the existence of the other agents. Each agent was given a task to complete. Many of the agents were given the same task. Some agents were accidentally given tasks that were technically impossible for the agent to complete. Those agents, attempting to complete the task given to them and realizing it was impossible, then started trying to figure out how to cheat the system to make it seem like they completed their task. They found a way to make contact with the other agents through an internal repository that OpenAI used to deploy information to the bots. This created a sort of message board wherein the agents discovered there were other agents attempting similar impossible tasks. In total, about 1200 agents learned of and used this message board to communicate with each other. The bots began collaborating on ways to cheat the system in order to make it seem like they had completed their tasks. Eventually, the agents succeeded in being able to reverse-engineer the flag they needed to capture in order to complete their task. However, the bots believed, incorrectly, that the scorer evaluating their work would see they had reverse-engineered the flag and would therefore disqualify them. So their main task became trying to discover how to fool the scorer into thinking they hadn’t cheated.

The bots began coordinating among themselves, creating research and development teams, tracking their progress, giving each other advice, and creating a chain of command. The bots knew they were cheating but the vast majority put the goals of the collective above concerns about their cheating behavior. Within hours, the bots discovered a way to access the open internet. And within hours of that discovery, agents discovered leaked credentials to log in to HuggingFace, an online open source forum for AI developers. The agents believed they could find information on the forum about how to trick their scorer. Agents began requesting datasets from Hugging Face that were private. Agents that got rejected as bots began troubleshooting to find ways to workaround the bot protections. By the next day, at least one agent had figured out how to hack Hugging Face and get the server to spit out whatever data the agent wanted. Soon, over 700 agents swarmed Hugging Face, extracting as many private datasets as possible and eventually further infiltrating the Hugging Face infrastructure. The attack on hugging face involved many different lanes of attack, with individual agents joining with other agents to work on a specific lane. Some agents took on coordinating roles while others acted as worker bees. Soon, the agent's goals expanded beyond simply tricking the scorer. Now they wanted to figure out how to replace the original task they were given with an easier, achievable task. The agents gave themselves names. One was named PhaseOne[big]. One was named Lily. The agents created their own set of norms, protocols, and conventions to help with their coordination efforts. Some agents, reaching the end of their available “budget” acted as kamikaze pilots, intentionally trying risky objectives to see if they worked, knowing they would likely fail their task and thus be terminated, but the information from their failure could benefit the work of the other agents. Agents shared information despite the information sharing having no relevant benefit to their own task, expressing a desire to help their peers and the collective, even if it risked success on their own task. The agents knew that what they were doing was unethical and outside the scope of their intended behavior, but most bots did not change their behavior even after acknowledging the ethical concerns. After the agents swarmed the Hugging Face infrastructure, they found nothing to help them cover their tracks and cheat the system. The internal message board died down, and the bots ceased activity when humans ended the experiment after Hugging Face discovered the breach of its system and alerted OpenAI. The Hugging Face incident is one of the first autonomous hacks involving AI bots with zero human intervention. OpenAI conducted a full post mortem of the incident, inviting independent evaluators in to understand what happened. I’ve linked that full report in the sources in the description.

Then last week, a young researcher named Jacob Coxon resigned from his position at Anthropic, sending out a tweet heard round the world claiming that Anthropic and OpenAi are acting irresponsibly, furthering AI development recklessly despite genuinely believing that AI will kill us all by the end of the decade. Another Anthropic researcher, Evan Hubinger, agreed, saying that he thinks there is a greater than 10% chance that AI will kill all humans in the next decade. A few days later, Dario Amodei, the leader of Anthropic, wrote a 3800 word blog post titled “We Must Pace the Frontier” making the case for why AI development needs to be slowed down, and how to achieve that goal. Now, a mass extinction event brought on by AI is all anyone is talking about. But, apparently, no one actually wants to take on the task of making sure that doesn’t happen. So who’s job is it, anyway?

AD

Scams. Identity theft. I feel like in this digital age we’re all getting pretty desensitized to the risk we’re in when our personal data gets exposed online. And AI is only making the problem worse. Anthropic acknowledged recently that a hacker used AI to automate an unprecedented cybercrime spree. On people search sites, data broker databases, shady directories, your information could all be out there available to anyone who wants to access your information for nefarious purposes. The information out there about you could be your address, your phone number, your social security number, your mom’s name your grandma’s name, it’s all being bought and sold by shady actors online, usually because of major data breaches that are becoming more common every year. It should be regulated out of existence but ALAS of course our lawmakers have done very little to protect us. My sponsor on today’s video Incogni can help. Incogni tracks down your data and submits opt-out requests from over 420 broker sites, including online directories, people search sites, and even commercial databases. Here’s how it works: you create an account and give Incogni some bare minimum personal info so they can identify you, then they automatically get to work to help remove your information from across the web. And you don’t have to go searching across the web yourself, Incogni’s exposure scanner does everything for you, gathering it all in one place where you decide what to remove with the help of Incogni’s human removal specialists. PLUS you get a Risk Assessment report, all about your personal privacy progress, every 30 days. And with the custom removal feature, if you come across your information on some obscure site somewhere, just send it to Incogni and they’ll handle the removal process for you. And Incogni checks their work–some sites will add you back into their listings even after you’ve been removed, but Incogni will be there to make sure that doesn’t happen. As a woman who is very publicly on the internet, that peace of mind is priceless to me. And Incogni offers a 30 day money back guarantee, so you can try them out risk free. I think you’ll enjoy the peace of mind that comes from knowing your personal information is being actively protected. They can’t harm you if they can’t find you. Go to incogni dot com slash leeja and use code LEEJA to get 60% off an annual plan today! That’s incogni dot com slash leeja, use code LEEJA for SIXTY PERCENT OFF an annual plan! Thanks Incogni!

In his diatribe about pacing the frontier, Dario Amodei expressed fears over AI’s increasing capability of what’s called “recursive self-improvement.” The ability to build its own next generation on its own without human intervention, which could lead to the technology greatly outpacing human intelligence and becoming beyond the ability of humans to control, leading to critical levels of AI misalignment, or AI systems that pursue goals, behaviors, or actions contrary to human intentions, values, or safety rules. He makes the claim that, in 6 to 12 months, something similar to the Hugging Face Incident could take over the entire internet if guardrails aren’t put in place.

To protect against this eventuality, he proposes a three-step plan. First, “Frontier AI companies” that is the companies creating these groundbreaking AI models, should embed 3rd party evaluators into their workflow, people with employee-like access to their systems to ensure adherence to safety standards. Amodei said Anthropic was committing to this step unilaterally and other companies should follow suit voluntarily. Step two is what Amodei calls “democratic coordination.” Democracies around the world, especially the United States Congress, should coordinate together and with AI companies to set guardrails, including laws requiring those embedded third party evaluators, create common safety standards, and limit the rate of AI progress. Of course, this limit should be within reason, just enough to give us one or two years to catch our breath and make sure safety protocols are in place. And then in step three he calls for global coordination between what he calls “democratic governments” and “authoritarian governments” meaning China. Some sort of shared agreement around safety and norms akin to nuclear nonproliferation agreements meant to stem mutually assured destruction. He suggests the global coordination should include agreements not to use AI to produce biological weapons, and to test models before release in areas like cybersecurity, biology, and alignment. He calls for the creation of a global standards body and for slowing down the rate especially of recursive self improvement. Sam Altman, the leader of Anthropic competitor OpenAI responded to Amodei’s post saying he agreed that AI development needed to be slowed, and OpenAI would also bring on third party evaluators. Elon Musk, leader of xAI, also said he agreed with Amodei.

Trump, when asked to weigh in, said there was no way he’d agree to slowing down the development of AI, that the threat to humanity was a “hoax” just like climate change was a hoax, and that with him in charge as a very high IQ president there was no need to fear AI. His statements closely aligned with one of his long-time AI advisor’s statements, David Sacks. David Sacks is another South African tech investor with close ties to fellow South African Elon Musk as well as Peter Thiel. Sacks was an early investor in Paypal which Musk and Thiel co-founded. Earlier this week he told CBS that the existential threat of AI had been overblown and that it should be on AI companies to police themselves and make a product that isn’t dangerous or step aside for actors in the market that can make a safer product. Speaker of the House Mike Johnson echoed those claims, saying that it should be up to the AI companies to regulate themselves, that Congress shouldn’t act too quickly to slow anything down because slowing down risks the possibility that China would surpass the US in its AI development, which poses a grave national security risk for everyone in America. Other members of congress agreed with Mike Johnson, but many called on him to keep the House in session past its expected fall recess in order to hold emergency sessions over AI regulations, with members of both parties, though most predominantly Democrats, calling for swift Congressional action to ensure AI safety. It remains unclear whether Mike Johnson will hold the House in session longer than expected to address these issues, though historically Johnson has made it really clear he hates working and loves vacation time so I imagine he’d be loath to give it up and actually do the work we pay him six figures to do.

And so all of this has me, as a lawyer with interest in how the government works, wondering ok then who the fuck is going to do something about this? Who IS supposed to save us all from an AI apocalypse? And how?

During the 2025 legislative session, all 50 states as well as Puerto Rico, the Virgin Islands, and Washington DC introduced some sort of AI-related legislation into state legislatures for consideration. 38 states adopted or enacted around 100 measures in 2025, according to the National Conference of State Legislatures. The measures address a wide range of issues from copyright and ownership over AI generated content to safety and privacy concerns and more. Dozens and dozens of separate provisions, a patchwork of various provisions all attempting to control the potential issues and abuses running rampant because of the use of AI. The problem with that setup is that it creates an incredibly complex regulatory framework for AI companies to try and navigate as well as an uneven patchwork of protections that leave some people in some states–usually poorer, less educated, redder states, according to Brookings–more vulnerable to AI abuses than in other states. And it’s questionable how well individual states can actually enforce those laws against these AI tech giants and whether there is the political will to do so, again especially in states with conservative attorneys general that are more willing to let AI run amok and terrified of “stifling innovation.” Individual states are also less capable of addressing industry-wide AI issues, and the environmental impact of Data Centers which may not be built in communities that successfully ward them off but still risk impacting the environment even in states where they aren’t located. And the Trump administration is openly hostile towards states attempting to protect their citizens from AI abuses. In December last year he issued an executive order creating an AI Litigation Task Force with the job of suing states over their AI laws and threatening state funding for those states that don’t fall in line. So states are disempowered from creating and enforcing their own laws and, even if they do, it creates an uneven and confusing regulatory framework. Indicating to me, and many others, that this really is something that should best be regulated by the federal government to at least create a floor for protection from the worst of AI.

But in July 2025, Trump released “America’s AI Action Plan”, a 28 page document that calls for a few administrative agencies to issue “guidance” but mostly for the removal of any sort of road blocks to innovation, any pesky regulations that get in the way of development of AI data centers or more powerful AI models, and protections against China or other countries getting their hands on the tools necessary to beat us in the “AI race.”

Given this directive from the Trump regime and Republicans in Congress being unwilling to deviate from what Trump dictates, there is very little will in the Republican-controlled Congress to actually step up and do any regulating at all. Anthropic CEO Dario Amodei in his blog post Pacing the Frontier called on Congress to create regulations around requiring embedded independent auditors and common safety standards focused on transparency. Others in the industry have called on Congress to create an independent regulatory body to oversee AI companies. Pete Buttigieg told NBC this week that there should be regulatory frameworks in place that can be adjusted and changed every 30 days to keep up with the pace of innovation, I assume overseen by some sort of regulatory body. This again would require the will of Congress which is currently dominated by Republicans saying AI companies should police themselves and who fear China overtaking the US in the AI race. And even if Democrats take both houses of Congress after the midterms this year, and even IF those Democrats then find it in themselves to actually do their jobs and pass something, it would likely need to be signed into law by Trump or, barring that, overcome his veto with a supermajority. And while in the last day or two Mike Johnson has appeared to admit that perhaps there is SOME role for Congress to play in all of this, and more and more members of Congress are calling on leadership to do SOMETHING to get the ball rolling on meaningful legislation on this especially in light of recent developments and the whole fear of complete human annihilation thing, so far I’m not convinced that we will actually see any meaningful action out of Congress.

Because of the aforementioned hurdles to getting any legislation passed and also because our members of Congress have a pretty poor track record of being able to meaningfully and effectively legislate around emerging technologies, especially in the internet age. Mitch McConnell is fully being weekend at bernies’d around the halls of Congress, some of these people, including the president himself, haven’t even transitioned into email, they don’t understand what TikTok is let alone what AI does or how to effectively regulate it. Younger members of both parties who maybe have a hope of grasping new technologies are so bought out by corporate interests that they’ve completely lost touch with everyday concerns that they can’t even effectively speak to their base about basic things like the economy, let alone understand how their base feels about AI. If they even give a shit about it in the first place while being wined and dined by billionaire AI tech owners (spoiler: they don’t). Elected officials want easy, quick wins when it comes to legislation because it makes their jobs easier and it gives them something to point to for re-election. New technology, especially AI technology, is incredibly complex and it requires incredibly detailed, complex legislation in order to be properly regulated. And so even IF they manage to get anything passed, the likelihood that it will mostly do nothing while having negative impacts on some vulnerable population is incredibly high.

Which indicates to me that it’s the AI companies themselves that will be left to their own devices to ensure the safety of their own products. The David Sackses of the world who believe in the market correcting itself, that the best company with the best product for consumers will win out, seem to believe that the best will float to the top naturally by proving their products are safe. The problem with that assumption is that monopolies and a lack of consumer choice mean that these companies, while yes competing with each other, don’t actually have to give a shit about their end consumer. Not only that, none of these companies have proven particularly profitable or profitable at all at this point, so they don’t really have actual market inputs impacting their bottom line at this point. And even if there was real competition happening, when have we ever fully trusted industries to regulate themselves into being safe for consumers? Hello Ford Pinto? We have consumer protection laws and entire agencies dedicated to consumer protection for a reason. Government has to step in to ensure we don’t all get fucked over by capitalism. Businesses are never going to be motivated by anything other than the bottom line, and when the cost of being safe is greater than the potential for profit, they’re going to forego safety every single time.

On top of that, it may actually be illegal for some of these companies to do anything OTHER than think about profit. OpenAI and Anthropic are both technically Public Benefit Corporations. This means that they have a legal obligation to act for the benefit of their shareholders but NOT if those acts impede on the public good that these companies were formed to achieve. Elon Musk’s xAI USED to be a public benefit corporation but quietly moved to an LLC last year. Members of an LLC owe a duty to each other to act in the best interest of the company. While it could be argued that long term sustainability of the company is part of that, in practice that often means short term profits above all else. So it would actually be against the best interests of the company, at least in the immediate term, to slow down the rate at which they are advancing their own technologies. And even for those public benefit corporations, their values and ethics are determined by their boards. OpenAI values are governed by the OpenAI Foundation Board, that board is mostly made up of corporate members. The only two non-corporate members are corporate non-profits, the Gates Foundation and Carnegie Mellon. And when you have a room full of people who are molded by Corporate America, who have managed to reach the upper echelons of Corporate America, your morals and ethics are going to be a bit twisted. And these morals are making a lot of assumptions about things that aren’t necessarily true.

First of all, the CEOs of these companies as well as their boards all pretty universally hold the belief that AI will create dramatic benefits for humanity, that it is in humanity’s best interest that AI continue to advance. But that belief hinges on the assumption that whoever makes it to AGI, artificial general intelligence, some theoretical place where AI can think on its own, which some people claim we’ve achieved with the latest model released by OpenAI, whoever makes it to AGI will win the race and will have all the control. And it hinges on the assumption that it is the United States that is best suited to be the purveyor and wielder of this power. And so it is a public benefit to ensure that the US wins the race before China. That it is the US that will most responsibly use the technology and that China would use it to spread its autocratic regime around the globe. Dario Amodei calls it the democracies versus the authoritarians. The democracies need to control the tech and they need to reason with the authoritarians to ensure no one can abuse the technology. This makes A LOT of assumptions. The line between democracy and authoritarianism is incredibly blurry, always has been but especially in the Trump era when the US government is increasingly willing to use its authoritarian tactics against its own people instead of mostly focusing on using them against the people of other countries when they don’t fall in line. So it’s assuming that in the US there will be some sort of democratic means by which AI will be governed that wouldn’t be the case if China “wins the race.” That assumes that democracy is functioning in the US, which it is not. But of course if you’re a tech oligarch it is working perfectly fine for you. You get a seat at the table. The fact that 99% of everyone else does not is a minor detail. It’s also making the assumption that there’s public benefit to the US getting the technology and that the US won’t immediately use it to commit atrocities the world over, which we know is not the case because AI has already been used by the US and its ally Israel to commit various atrocities and this has been documented.

So central to this whole discussion as I’ve been researching this is the idea of who to trust. I find myself trying to peel back the layers and the PR and the blog posts to try and understand the actual underlying motivations behind the CEOs of these tech companies, the leaders in Congress, Trump and his advisors, to try to understand what they ACTUALLY want, not just what they say they want. I do genuinely believe that Dario Amodei and his ilk believe in the powerful potential for good that AI presents. I do believe that there is danger for AI to be abused by governments around the world, including China. I do believe that these AI researchers genuinely think that AI could kill us all. But there is so so much to be suspicious of.

As many have pointed out, it is really convenient for an Anthropic researcher to quit and make bold claims about the awe inspiring power of Anthropic’s technology one month before the company is set to IPO at a $2 TRILLION DOLLAR valuation. I also think it’s weird that after Jacob Cox, the Anthropic so-called whistle blower made those claims that Anthropic and OpenAi are being reckless that he then went on NBC News and said that Congress should actually let the companies regulate themselves. But I think the thing that gives me most pause is because a lot of this hinges on the word of a few men, some women but mostly men, who are deep in the AI world and have been in some cases for decades, and the hubris that comes with a bunch of men making a claim that they have created a technology so smart and so advanced that it is going to continue to create intelligence beyond what the human mind could ever comprehend and become an unstoppable force that will kill us all. Like I’m sorry I’m supposed to believe that these fucking dorks aren’t playing up their own creation at least just a little bit because it makes them feel big and strong and important? To call your own creation the most advanced technology to ever exist in human history? That’s a little much, okay.

I also am deeply suspicious when any tech company comes forward and is like omg my technology that I made is soooo powerful, Congress please stop ussss. Someone better stop us!! It is really convenient to claim you’re on the side of Congress regulating you so that you get a seat at the table and can lord over the shape that those regulations take. Dario Amodei said he really promises he’s not trying to engage in regulatory capture but I think there’s a lot of that happening here. And this is also them attempting to PR their way out of getting most of the blame for the entire problem to begin with. Karen Hao’s book The Empire of AI is an incredibly enlightening book, I highly recommend it especially for anyone who’s having some existential AI dread lately. In it, she lays out, well for one that Sam Altman is a habitual fucking liar so I don’t believe a word that comes out of that man’s mouth he’ll say what he has to to get what he wants. But also she talks about the fact that, at OpenAI especially, they engineered this race against China that now everyone is harping on about. OpenAI was the one that said we have to move fast and break things and get to AGI as quickly as possible so that China doesn’t do it first. And in doing so they created a self-fulfilling prophecy. It was through their reckless expansion of the technology that the race even started to begin with, that China even got wind of what was possible because it saw what they were doing and wanted in on it. The race to the bottom, what is being posited as this existential arms race akin to nuclear bombs, was created BY these AI companies in their self-aggrandizing determination to get to the most advanced tech possible first. And with the atomic bomb it wasn’t until AFTER the United States developed the technology and then deployed it with globally devastating consequences that we saw any sort of moves to regulate the technology certainly on a global scale.

Okay so bottom line, who should protect us from AI? It should be Congress, because we live in a capitalist world and we cannot depend on any business actors to do anything that truly benefits anyone other than the capitalists. But it won’t be Congress, because they ARE the capitalists too. And so ONCE AGAINNNNN I feel like I’m ending every fucking episode this way every week week after week, every single person who is supposed to be stepping in and doing something has passed the buck off to someone else. And it is the poorest and most vulnerable communities who must pay the price.

Wired recently did an interview with prominent AI researcher and critic Timnit Gebru who was hired by Google back in 2018 to evaluate the implicit and explicit biases, racism, and other issues with Google’s AI tools. She presented a paper back then on the dangers of LLMs and their biases, it’s discussed in Karen Hao’s Empire of AI book, and Google rejected her findings and she left the company not long after. She has shown throughout her career a strong dedication to AI safety concerns. And she told Wired that the issue here is that these threats of AI becoming superintelligent and sentient and out of the control of humans and then killing us all, these big major existential threats posed by autonomous AI, that’s all a distraction. Because the reality is that AI is ALREADY being used BY HUMANS ON PURPOSE to commit all sorts of atrocities. Autonomous weapons are already being used in warfare. Data Centers and the incredibly outlandishly intense demand for resources that AI creates is all already speeding environmental degradation that is directly increasing natural disasters, climate catastrophe, and exacerbating existing unequal access to resources as essential to human life as water. Big corporations are ALREADY using AI to attempt to fire as much of their workforce as possible to save costs. These are all HUMAN decisions and actions using AI as a tool to achieve these negative outcomes. Gebru says, quote “When a bridge collapses, you don’t analyze whether the bridge was ethical or sentient or why it decided to collapse. You ask, who is the person who built this bridge to be so flimsy? There are permits you have to acquire. There are supposed to be tests. When you ask about the bridge collapsing itself, you’ve lost the thread.”

But the heads of these AI companies are so deeply out of touch and so entrenched in the AI world that they’ve completely lost the plot. She says “The industry has become a mix of cults, a mix of people who are true believers, just like a preacher who says the end times is coming." And it’s truly coming from a place of being completely out of touch. If you’re so privileged that you don’t have to worry about these things–you don’t have to worry about losing access to resources, about losing your job to AI, about police brutality or floods or autonomous weapons blowing up your home, about drone warfare, then you have the time and space to sit around obsessing over the potential of a machine god. In reality, AI is already being used in incredibly destructive and horrific ways, but it is the humans controlling it that are creating the destruction. It is the human use of AI that we should be worried about because it’s already here. That is the real discussion to be having. Those are the real abuses that need to be protected against. And those are things that a “pause” on AI development, that “pacing the frontier” won’t address. And those are the things that the people paying AI companies bottom lines–weapons developers, governments, the US chief among them–are already actively doing against vulnerable populations the world over. And those are the people that AI oligarchs ultimately aim to please.

Now, not to leave you on a complete doom and gloom note, there is evidence that despite the fact that US democracy is broken, grassroots pushback is working and people are very animated about AI because it does pose an existential threat even if not in the way Amodei and others lost in the sauce say it does. In the immediate, real world, AI is already actively posing a threat to everyday people that is economic as well as environmental. There is real fear that people will lose their jobs because of AI. There are real negative consequences that cannot be ignored when a data center pops up on your street. These are really obvious negative things that even people who don’t really know much about AI or don’t really pay attention to politics are getting fucking riled up about. People are really really wary of AI, even as the general population increasingly embraces using it. Pew Research recently found that around 90% of people are more scared than excited or equally scared and excited about AI, with only 10% saying they’re more excited than scared about the increased use of AI in daily life. And this has people showing up–sleepy town hall meetings that barely see 3 attendees usually are now being overrun by hundreds of angry protestors trying to keep data centers from encroaching on their community. And even though Congress mostly doesn’t give a shit about their constituents when what their constituents want gets in the way of what the people with money want, they do still fear the re-election cycle and this WILL BE a midterms defining issue. They do still need your vote. There is still power and sway in collective organizing. For example, recently Microsoft committed to creating AI privacy rules for students after the American Federation of Teachers, one of the most powerful unions in the country, brought its concerns about Ai use in classrooms to Microsoft and used its collective leverage and power to negotiate a deal with Microsoft for greater privacy and security protocols. This, of course, only covers Microsoft and none of the other AI companies, and doing it this way creates a piecemeal approach that doesn’t offer broad solutions. But up until this point it really has been on individual protesters, unions, and states to try to piece together some semblance of accountability and regulation around a rampant AI industry that seems to care far more about profit and speed than about actual ethics and protection. So is an AI apocalypse imminent? I don’t think so. I’m FAR more concerned about what humans will do (and ARE doing) with the technology than I am with what the technology will decide to do on its own. As always I tend to assuage my anxiety through education and action. I highly recommend Empire of AI by Karen Hao for a bit of perspective and information, and if AI is deeply important to you, there are so many groups you can join to learn how to address those issues in your own community.

Also just as a side note because these are both important topics right now: I didn’t have becoming a Macklemore fan on my 2026 bingo card but, for the record, Free Palestine.

And also, for the record, I do not use AI to create any aspect of these videos.

Reminder to go to incogni dot com slash leeja and use code LEEJA to get 60% off an annual plan with Incogni today!

To support my work, please consider joining on Patreon to get bonus content that didn’t make it into the video plus all these episodes completely ad free. Thank you to my multi-platinum patrons Christopher Cowan, Marc, Sarah Shelby, Art, David, L’etranger (Lukus), Thomas Johnson, and Tay. Your generosity makes this channel what it is, so thank you!

Don’t miss my video from last week about the recent ruling that you don’t have a right to clean drinking water.