The 404 Media Podcast (Premium Feed)

from 404 Media

Inside Musk’s Government Takeover

You last listened February 6, 2025

Episode Notes

/

Transcript

We dedicate the whole free section of this episode to Musk’s takeover of the federal government. We got audio from a meeting in which a Musk ally laid out their plans for AI across the government, and revealed who is now the head of HR for DOGE. In the subscribers only section, we talk about how your public library probably already has a bunch of AI slop.

YouTube version: https://youtu.be/364fgAuALD0
Joseph:

Hello, and welcome to the four zero four Media Podcast, where we bring you unparalleled access to hidden worlds, both online and IRL. Four zero four Media is a journalist founded company and needs your support. To subscribe, go to 404media.co, as well as bonus content every single week. Subscribers also get access to additional episodes where we respond to their best comments, gain access to that content at 404media.co. I'm your host, Joseph.

Joseph:

And with me are two of the four zero four Media cofounders. The first being Emanuel Mayberg. Hello. And then Jason Kebler.

Jason Koebler:

Hello. Hello. Good to be here.

Joseph:

Yeah. So we're gonna be talking about Musk and Trump and the US government, because it is an incredibly important series of stories and coverage right now. We can do it in a little bit different way. We're gonna talk about that, then talk about two of our stories spread out a little bit. Jason, what did you wanna, like, sort of explain at the top?

Jason Koebler:

Yeah. I mean, I wanted to say that when Trump was elected, we did an article about how we were gonna cover the Trump administration in the same way that we've covered previous administrations because what the government does is very important, especially when there are so many, like, gigantic tech companies who are involved in this administration in some way. And I think one thing that we said is we don't want to, write only about politics and Trump and the government because there's a lot of other things going on. There's also a lot of people reporting on politics at the moment. And so there's a lot of, like, civilian stories that are happening, like, you know, private companies doing things that we are working on and that we wanna talk about and that we wanna cover.

Jason Koebler:

I think all of that said, the outlets that are really rising to the occasion right now are wired. I think we've done really good work and a few other, like, tech focused publications. And one of the reasons for that is because this administration is obsessed with running the government much like a tech company, and a lot of the people involved in it are either former, current Tesla employees, Neuralink employees, you know, associated with Elon Musk in some way, or they're like Silicon Valley Venture Capitalists. And so that's to say, if you don't like, you know, politics in general and if you want to hear about something else, this is not gonna be that episode, but we are not going to cover only this. It's like our focus as a publication is not changing in the long term.

Jason Koebler:

I think it's just in the short term, there's so much happening right now. We have very important sources who have put themselves at risk to talk to us. And so, you know, we're trying to report on what's going on, and that takes a lot of our attention and resources. And, frankly, like, a lot of people are paying attention to what is happening right now, and so we don't wanna publish a ton of stories about unrelated things, have no one pay attention to them, and then, you know, them not getting the attention that they deserve. I mean, do do you do y'all think that's fair?

Jason Koebler:

Do did we even need to say that?

Joseph:

I don't think we needed to, but I think it's a fun sort of conversation, and it helps people contextualize and understand. I mean, I would just stress that, you know, we are primarily a technology publication. We cover tech companies. We cover surveillance, privacy, sex on the Internet, how all of these things, intersect as well. But I would say, like, this election and this administration is probably the most techified administration it's ever been.

Joseph:

You have AI and the push from there, and we'll get to that too, I think, our first story where we're talking about this push to have more AI in the in the US government. And that will have impacts inside government, but also, you know, just ordinary people's lives. You know? And then, of course, as you say, you have the Musk connection as well, the richest man in the world and the owner of multiple tech companies. Like, it's no longer, it's no longer the time where people can say, oh, you should stick to tech.

Joseph:

I think that was stupid when people say and nobody said that to us, to be clear. I'm only even thinking, like, way back to 2014 or something. You know what I mean? Or, even defective being told, stick to sports or something. No.

Joseph:

No one's saying that to us. All I'm trying to get across is that there is a massive amount of tech going on here, and it's really, really gonna impact people. Emmanuel, what do you think about that?

Emanuel:

I think that if you don't wanna hear about politics, you should become a paid subscriber and listen to our bonus section about libraries and AI.

Joseph:

That's actually a very fair point. Yeah.

Jason Koebler:

I mean, the story that Emmanuel published today is, like, fantastic reporting on AI, which libraries there's poll there is a political angle there too, which I'm sure we'll bring up. But, yes, that was a fantastic investigation, and people are reading it. So, obviously, people care about other things, that are going on as well, but most of the free section of this podcast is gonna be about Musk's takeover of the federal government.

Joseph:

Yeah. And before we talk about our story that we just published based on leaked audio that we'll get into, let's just do a really quick rundown of what is happening or what has happened recently. Because honestly, I'm losing track. We were writing that article, and, it's just hard to remember what happened from one day to the other. So, I mean, there's a few things, like, Musk's allies have gotten access to sensitive treasury data.

Joseph:

Right? Gained access to classified information related to USAID, probably gonna shut down the consumer financial protection bureau, which is an agency we covered recently because they've started taking action against some of the companies we cover, you know, like privacy and surveillance and all of that. Some of the people associated with Musk's Doge agency and then the agencies around that, are very young engineers or coders from Musk's other companies. Jason, did I miss anything?

Jason Koebler:

I mean, you missed the you missed the government websites going down and the archiving of information, which we did talk about a little bit last week, but there's sort of, like, wholesale deletion of all of these big, US data databases. Anything having to do with DEI broadly defined, a lot of climate stuff is coming down. And then we've covered a lot of the efforts to archive that. I did a big piece about where that stuff continues to live on that, you know, you should go check out. But there's sort of the the dual tracks here.

Jason Koebler:

There's, like, the Musk track and the, like, government website purge of government data, which we're actually more familiar with because that happened during the first administration as well.

Joseph:

And you covered it then for motherboards as well. Right?

Jason Koebler:

Yeah. Yeah. Whereas, you know, this sort of, like, let's bring on Silicon Valley to start up ify the and automate the government is pretty new. I mean, conservatives have always wanted to cut government programs and government funding and say, you know, the government is too big, things like that. But I think that the techification of it and the sort of, like, let's use automated systems and AI to do it is is fully that's new as far as I know.

Joseph:

Yeah. That sounds about right. Alright. How about we leave that intro there, and then we'll have a quick break. And then when we come back, we're gonna talk about one of the specific stories we just published about how a Musk ally is trying to push AI, across the US government.

Joseph:

We'll be right back after this. Alright. And we are back. Here's the story that we all worked on, actually, all three of us. It was kind of a scramble.

Joseph:

The headline is, things are going to get intense, how a Musk ally plans to push AI on the government. Jason, I said we all worked on it, and that is true. But you did most of the heavy lifting here, to be clear. Me me and Emmanuel Moore just helped out, you know, and and contextualized that sort of thing. I guess, first of all, what was this meeting?

Joseph:

Who was involved? Who were they talking to? And then we'll get into what was said.

Jason Koebler:

Yeah. So maybe let's start with the agencies that are involved, which I'll try to do as as quick as I can. So there's DOGE, which the Department of Government Efficiency, or it's called something else now to make it, align with a specific acronym. But there's basically DOGE, which is Elon Musk's thing. And the way that that agency was created was on day one, Donald Trump issued an executive order establishing Doge.

Jason Koebler:

And the way that he did it was he took this agency called the US Digital Services, which falls under the, office of man budget and manage OMB. Office of management and budget. There's so many acronyms, and remembering all of them is tricky.

Joseph:

Crazy, man. It's crazy.

Jason Koebler:

So it fell under OMB, and USDS was formed in the aftermath of the healthcare.gov debacle. We actually did talk about 18 f a little bit last week, I think, but, basically, USDS is this digital services, department of coders, more or less, who hop between agencies and solve problems. And Trump renamed that, put Elon Musk in charge of it, and put it directly underneath the president.

Joseph:

Let me just clarify one thing. The reason they renamed, the United States Digital Services is because he can't create a new agency whole cloth, so they had to basically rebrand one. Is that right? That's my

Jason Koebler:

That's that's right. And he picked a specific one that falls under this, like, special authority that I'm not gonna get into right now because it's way too complicated, and I think I do understand it, actually. But but, basically, it it is an agency that the president has a lot of control over. And he put it directly underneath his chief of staff, which means that it's not beholden to specific, like, hiring requirements, which is why you have, like, random people working there more or less. You know, it has different record keeping rules, which we'll probably do more reporting on.

Jason Koebler:

But, basically, like, there's Elon Musk and there's Doge. And then Elon Musk's Doge people have been going into different agencies. Like, they have been essentially saying, like, under our executive authority, under the president, we are going to come help you cut costs. And what that means is we're gonna help you, like, gut agencies. So they've gone into USAID, which is the aid organization.

Jason Koebler:

They've gone into treasury, as you mentioned. And they've also gone into this agency called GSA, which is the General Services Administration. We did talk about GSA last week. But they basically do, like, a lot of, coding for other parts of the government. They do a lot of standard setting for other parts of the government, and they also are in charge of, like, government real estate and things like that.

Jason Koebler:

So under GSA, there is this agency called TTS, which is something technological services.

Joseph:

Good.

Jason Koebler:

Yeah. So there's TTS. And then under TTS, there's something called 18 f. And 18 f It's

Joseph:

the it's technology transformation services. And, of course, we know that. It's just that when you ask to list all the agencies and the acronyms, it's fucking confusing.

Jason Koebler:

Yeah. And the reason that I don't have all of this in front of me right now is because we've literally been reporting around the clock constantly. Like, normally, we we prepare more for this podcast, so please forgive us because it's like, we just published this story. Our hair is on fire. We're running around like crazy.

Jason Koebler:

So that's a huge preamble, but, basically, there's this department called TTS. And the person in charge of this is a former Tesla engineer named Thomas Shedd. And so he's basically, like, a a Musk plant at this agency. And so even though he's not technically part of Doge, he is, like, an Elon Musk ally. They're working very closely with Doge, and he oversees this group of hundreds of government coders.

Jason Koebler:

And so

Joseph:

They're basically government software engineers in a way, and they're like product managers, and they they build, products for the government, basically. And I know the word product is kinda strange to some people when you're talking about government, but that's also in the same sort of terminology that, you know, intelligence agencies use and that sort of thing. We're developing a product for x y z or something.

Jason Koebler:

Right. So Thomas Shedd had a a big meeting with these government software engineers, and we got audio of that meeting. And in that meeting, he talks about what he wants them to do, more or less. And this includes things like doing changes to login.gov, which is one of those products that that Joseph just mentioned that the government works on where it's a unified login system across, like, all of the US government. So, like, I have a login.gov account that I use to do things with such as, like, TSA precheck and things like that.

Jason Koebler:

He basically said I think it's important to read the quote more or less, but he said, you know, I'm gonna need help with changes to login.gov. And one thing that he said was he wants to start using it as a mechanism to identify fraud and abuse within the government, like, identify US citizens who are defrauding the government in some way. And right now, login.gov doesn't have access to, like, any sort of government information across agencies. It's just like a login platform.

Joseph:

It's just a way to authenticate in some ways to to log in where, as you say, for TSA or whatever or for the IRS or or whatever it is, you don't need to make a different account for every single one, which is a massive pain, I imagine. So you just have this one account, but the he wants to add something to that.

Jason Koebler:

Yeah. So, like, you can't if you if you're on Social Security, if you're getting Social Security payments, you can use login.gov to log in to Social Security, like, Social Security Administration to see your things. But login.gov doesn't have access to how much money you're getting from the Social Security Administration, whether, you know, you have another account, like, stuff like that. It doesn't have access to your personal information, and he wants to change that. So he said, quote, in this meeting with the, with his employees, he said, quote, just like a fun one we've been thinking through with login.gov, specifically in TTS, is login can't access government information on individuals, and so there's no connection that login has with Social Security or any other government system even though we're part of the government.

Jason Koebler:

And so part of one of the things to work through now is how do we make it so that those agencies that have information of very secure APIs that can be leveraged by login to further identify individuals and detect and prevent fraud. To summarize, it's like, how can we use login.gov to detect fraud and identify the individuals who are committing that fraud? And then one of his employees on this meeting says, well, that that would be illegal because there's a privacy act that says, you know, these government agencies can't, just, like, share information willy nilly.

Joseph:

Yeah. You can't just have one agency randomly one day start sharing Social Security information with another agency for the hell of it, basically. And there would need to be some sort of opt in, consent. There there's actually that quote, from the employee question. What what was that quote exactly?

Jason Koebler:

He said, I think we were on the talk I think we were on the topic of a login aggregating data. It's an illegal task. The privacy act forbids agencies sharing personal information without consent. And then Thomas Shedd says, the idea would be that folks would give consent to help with the login flow. But, again, that's an example of something that we have a vision that needs to be worked on and needs clarified.

Jason Koebler:

And if we hit a roadblock, then we hit a roadblock, but we still should push forward and see what we can do. Emmanuel, I think you have some, like, a good insight on, you know, what he's saying here and why this is, you know, notable.

Emanuel:

I think the notable thing about it is that he makes very clear that they don't intend to break the law. Right? He's not getting on the meeting and saying, here are the laws we have to break. In order to do this, please break the law. He's responding to this question and saying, okay.

Emanuel:

I understand. We understand the legal limitations of what we're trying to do, but there is a vision. Right? And they're striving towards this vision, and they sort of say, I don't know if we ended up quoting this, but at some point, he even says we'll cross that bridge when we get to it, which I think just tells you a lot about the mentality of the Musk camp in the government and what they're trying to do. They are looking at the rules, and they're trying to navigate them to fit this vision of a startup like streamlined organization with a ton of efficiencies.

Emanuel:

And we happen to be very lucky to have seen Musk do this very recently at Twitter, and I'm not saying that we know that it's gonna look exactly the same, but we just saw him do this exact thing at a private company, and it gets pretty ugly. A lot of people lose their jobs. A lot of services go offline. The quality of services generally deteriorates, and the quality of their product overall is worse and less valuable. Right?

Emanuel:

So

Jason Koebler:

Yeah. I mean, that that's partially true. But the other thing that seems to be happening here is they're saying, we have a vision, and we're gonna start doing stuff. If it happens to be illegal, like, try and stop us, or we'll figure it out. Yeah.

Jason Koebler:

I mean, in some cases, I think the attitude is try and stop us. I'm not saying Ched has that ethos in this because he is he's saying, like, oh, we'll try to navigate it. But Musk, certainly, on Twitter is like, we're just doing stuff. We're gutting stuff, and we're getting access to the treasury, and we're doing whatever. And, you know, there's many lawsuits that have been fired filed, but it's unclear which ones are gonna stick, which ones like, what what actions are gonna be taken and and whether they'll be stopped.

Emanuel:

Yeah. And that's a mode of governance that is bigger than Musk at the moment. It is representative of what Trump is doing across the entire government. Right? It's not just Musk that is laying down a goal that seems unfeasible and says we'll figure it out.

Emanuel:

That is exactly what Trump is doing with the Department of Education, with the Department of Labor, with, I don't know, thinking Canada, state, whatever it may be.

Joseph:

So how does AI and automation come into this? Because Shaz talks a fair bit about that, especially at the top of the meeting.

Jason Koebler:

Yeah. I mean, he talks about a lot of things, and we had audio of the whole meeting. So one thing he talks about is having a team of coders. Like, he basically is like, I need volunteers to help me build AI coding agents, meaning, essentially, like, AI government employees that will write software for other government agencies, which is not I mean, Mark Zuckerberg was literally talking about trying to do this at Facebook, at Meta, in an all hands last week. And so this is a Silicon Valley dream where it's like, we're gonna have coders teach the AI how to write code, and then an individual human coder can be a hundred times more efficient, or we can just, like, deploy these AI coders to do a whole bunch of stuff.

Jason Koebler:

And then, you know, at some point, maybe there's, like, human oversight to say this code is good or this code is bad. I talked to one employee from TTS who is a software developer, and they were horrified by this proposal because writing government software is different from writing enterprise software in many cases. There's many more privacy laws. There's many more, you know, you're designing things to make sure that they won't break because a lot of times the software deals with things like Social Security payments or or things that you, like, really need to not go down, which is part of why a lot of government systems run on extremely old technology. So there's that.

Jason Koebler:

They also talk about trying to train some of these AI on or using AI to analyze government contracts to determine fraud, abuse, du duplicate contracts, things like that. You know, that that could also be problematic where you just take an AI and say, go analyze all this stuff and then make

Joseph:

decisions based

Jason Koebler:

on that.

Joseph:

Yeah. Ends up killing some contract that's actually really important or something.

Emanuel:

You

Joseph:

know? Like, we we look at a lot of US Government contracts. I'm always going through procurement databases. And I always FOIA emails often from GSA because they're the ones sometimes touching the the purchase of the product. And, I mean, I don't know.

Joseph:

They're very professional there. You know? And you go through these emails and they they're keeping track of stuff. And I've seen people be like, oh, we're still owed to this and that sort of thing. Obviously, that is anecdotal and just a snapshot, and not an entire view of the US government.

Joseph:

But you you sort of brought up motivations, and that was what really stuck out to me when I was listening to this audio and what I pulled out, which is that Sheds, again, the Muscally, really framed it almost like a private company trying to please shareholders. At one point, he says, you know, we're basically bankrupt as an agency, TTS. I believe an employee then says a question which is read out, which is something like, I don't agree with the bankruptcy claim. And then Shed replies, look, TTS is losing something like $200,000,000. If it was a private company, our shareholders would be telling us to do x y z.

Joseph:

And that is just so illuminating to me because the government is not a private business. The government is not there to make money. The government is there to provide services and security and various other things to its citizens. It's not there to make a buck. Like, what do you think that's

Jason Koebler:

It's literally like a it is a government service. And people make that argument about the post office sometimes where they're like, the post office loses all this money. And it's like, well, first of all, the post office doesn't lose that much money. They're we won't get into it. It's very it's a very well run organization, but it's providing a a service for millions of Americans across, you know, the entire country.

Jason Koebler:

And so the people who act like you can just, like, kill all this stuff and it having no impact is is really wild. I think, there was another really illuminating question on there where an employee said, we're not allowed to work more than forty hours a week with you know, because overtime rules and things like that. And he's basically like, yeah. We're working on that. We're working on that.

Joseph:

Consult HR policies, I think, was also mentioned as well, which is, like, you're just turning the government from a collection of civil servants into a fucking hackathon, where everybody's, like, gonna gather around and, like, try to churn stuff out. And, like, I don't know. That that can work for private companies and start ups. It's just the motivations are fundamentally different when we're talking about, government. You know?

Joseph:

Alright. Let's leave that there. We'll have a quick break. And then we come back, we're gonna talk about the second story we did about Doge and and Musk and that sort of thing. We'll be right back after this.

Joseph:

Alright. And we are back. This one, that me and Jason did. The headline is Musk's Doge brings in HR consultant focused on non woke DEI aligned with our faith. Obviously, the non woke and the aligned with our faith are quotes.

Joseph:

I guess, who is this person? Stephanie Holmes, I believe, Jason.

Jason Koebler:

Yeah. Stephanie Holmes is an HR lawyer who created a consultancy firm called Brighterside HR. And by her own explanation in these videos we went and watched, she created it as a non woke version of HR. She said basically, like, leftist ideals were infiltrating corporate America and were, you know, pushing diversity, equity, inclusion. And so she was like, I'm gonna teach corporations how to play that game more or less.

Jason Koebler:

She calls it a a game at some point and, like, redefine what DEI is to better align with the Christian faith is what she says in in one of these videos. And she's become this kind of, star on the right where she shows up at, like, heritage foundation events who, you know, they did project 2025, the Federalist Society, which is another conservative think tank, and then this group called the Napa Institute, which is a Catholic organization in California that is it's like a religious organization that funds conservative causes more or less. And she, for years, has been going to these events and basically saying, like, I am teaching companies how to be non woke more or less. And she was brought into Doge to be the head of HR for Doge more or less.

Joseph:

Whatever that means exactly. I mean, one of her quotes so just to clarify, we learn that she's appointed as the head of HR. We then go away and we look at other talking speaking engagements she's done, because, you know, that gives some insight into who is now involved with this agency. One of the quotes is, you know, I do workplace training, discrimination, harassment training, how to do workplace investigations. It's simply just a non woke version offering employers an alternative approach to diversity inclusion.

Joseph:

I mean, may maybe you don't even know, Jason, and maybe it's just me not being able to pass this sort of language. And I think you're in the manual a lot better at that than me. What does it even mean to be, like, non woke HR? Like, what does that mean?

Jason Koebler:

Well, HR is not super woke in my experience. HR is like

Joseph:

Corporate lackeys protecting the company.

Jason Koebler:

Yeah. Maybe does some, like, nods at caring about diversity and and things like that. But, like, by and large, HR at companies that I'm aware of do things like help cover up employee harassment and things like that. You know, not always, but that that's part of what they've they've done. But, basically, like, Stephanie Holmes, like, in these videos that we found, she said that very often employees are asking companies to diversify their workforce, which is true.

Jason Koebler:

Like, very often the push to hire more people of color or to, like, have a a more inclusive workforce comes from employees themselves, who have seen, you know, like, hey. Everyone who works here is a white male. Like, let's diversify in some way, which for many, many reasons leads to better, corporate outcomes. But she was basically, like, if you are an HR person or, like, a CEO who is under attack from your woke employees asking you to hire more black people, I will teach you how to, like, implement DEI programs that, favor diversity of thought, which is a code word that, you know, the right has been using for many years, that basically

Joseph:

said, like The Yeah. Yeah. That's why I find it so difficult. And, again, you and the manual are so much better because it's all code words. You know?

Joseph:

And I don't follow the right close enough to be able to translate what they're saying, basically. I mean, just to elaborate a bit, she says she advises employees to move away from divine defining diversity exclusively focused on employees' race, sex, or other protected category, and instead bringing together employees with diverse backgrounds, viewpoints, perspectives, and beliefs to achieve common workplace goals. I mean, on the face of it, that sounds good, but it it's a code word. Is this is is this what you're saying?

Jason Koebler:

Yeah. I mean, for people who really care, maybe it goes back even further than this. But at Motherboard, we were the first people to report on the Google memo, which was this memo written by a software engineer named James Damore at Google, like, back in 2017 or 2018. That was about Google attempting to hire more women software engineers, and he was arguing that there were, like, these biological differences between men and women, and women don't wanna become software engineers, blah blah blah. Like, that's what he was arguing.

Jason Koebler:

And he said, we have this goal of diversity at Google where, you know, we'll hire more people of color, we'll hire more, trans people, we'll hire more women to diversify our workforce. And this is all a means to, like, check a box saying that you have a more diverse workforce. But what we should really be going for is diversity of thought, meaning he was arguing that Google had become a monoculture, like, a a leftist progressive monoculture that was left leaning more or less. And so he was saying, like, you need to hire more conservatives more or less. And this idea of diversity of thought has become something that conservatives have said, like, institutions need to do more of.

Jason Koebler:

So when, like, The New York Times hires, like, a Republican to say that Donald Trump is good actually on its, you know, opinion pages, that's like the New York Times seeking a diversity of thought, I believe. And, it may it may predate James Daymore, but that was the first time I was exposed to it. And when we wrote about that memo, it became like a huge it became really like the beginning of the culture war in Silicon Valley. And I think that this, idea that companies need to hire more conservative people is, like, that's what they mean by diversity of thought. Like, that's literally what they mean is, like, we need you need to hire more, like, religious people, more conservatives, more people from rural areas, so on and so forth.

Jason Koebler:

I mean, Emmanuel, you've edited many, many stories about this sort of thing. Do do you think that's right?

Emanuel:

Yeah. It just it's a it's a pivot where you simply argue that rather than your workplace making an effort to hire more people of color, you should hire more Republicans. I mean, that it's it's simple as that. Or I think in her case, in Stephanie's case, more religious people or have more respect for Christianity in your workplace. That that's just kind of the the the pivot that she's making and all these people are making when they talk about diversity of thought.

Jason Koebler:

So I know that's a a lot about Stephanie's background. The reason that this is important, why anyone should care, is because she is now in charge of HR at Doge, which is this agency that's tasked with gutting the federal government. Like, that's its its specific, reason for existing. And so someone who's thinking through that lens is going to go into other agencies presumably or or could, and have an impact on, you know, the federal workforce. I think the other thing that was very interesting to me is we've been talking about how Doge and different parts of the federal government have been taken over by Silicon Valley types and Elon Musk's, you know, proteges and people who work at his companies and how like like, all these people who worked at Neuralink and SpaceX and Tesla are part of the government.

Jason Koebler:

Well, as far as I know, Stephanie doesn't have any connection to any of Elon Musk's companies, and yet she is part of this agency that Elon Musk is running, which suggests that there is some aspect to the Department of Government efficiency that is an alliance between Elon Musk and Silicon Valley and this, like, Christian right, which that that's me. I don't know exactly what the relationship is, but it's like she's essentially like a heritage foundation, Federalist Society type person, and she is now at this agency that is like a Silicon Valley type agency. And that's very interesting to me.

Joseph:

Yeah. And I think just to stress that, I think some people might be like, oh, why do we care about this person being involved with this? Or it also goes to, you know, Wired's, coverage of, you know, the young Musk aids and associates from other companies coming out. Like, well, why do we care about the individual people? We the individual people.

Joseph:

We should care about policies and stuff. Well, it's important to know who these people are because they are directly gonna impact policy, which is directly gonna impact a ton of other people's lives, both government workers and, you know, civilians as well. Right? So I imagine we'll if not us, you know, or probably including us as well, but other outlets will keep reporting on, well, now this person's involved, this person's involved. Sort of like when more straightforward political outlets will do, well, this person's been nominated for the head of the FBI or something like that.

Joseph:

Like, we wouldn't probably cover that as much, unless there was some sort of tackle surveillance thing or something like that. But here, I think the through line to the Google memos you say does pull it in a much broader context, and that's why people should care. You know?

Emanuel:

I'm sorry. I think people should also care because of the content of what is being said and done in these meetings that we talked about, in the previous segment. Right? Like, there's been a lot of ink spilled over the fact that these guys are young and are from Elon Musk's tech companies, which are both certainly interesting and notable facts. But I don't think any of us are thinking no one who is 22 should work in the government.

Emanuel:

You know? Like, if, as the as as the AI enthusiasts like to say on Twitter these days, if you get a a group of cracked 22 year olds that are really good at their jobs and they're working in the government, that's a win for the government, and I tend to agree with that. But, again, the issue and the thing that we are trying to highlight and the thing that the public needs to be aware of is that there is a vision and there is a goal. And, it is very new and untested and concerning given everything we know about AI and everything we know about the revolving door between government and private companies and now having, you know, this titan of industry and tech running seemingly large parts of the government. That is what is notable.

Emanuel:

I don't care that they're 22. That that is not a

Jason Koebler:

Well, some something about that. Like, I lived in Washington, DC when I was 22, fresh out of college, and I knew many, many 22 year olds who worked for the federal government. And the way that the federal government works by and large is they have very strict bands. They have, like, different levels, and I don't remember the exact way that it works. But it's like, you start at level one, and you go to, like, an intern.

Jason Koebler:

You're not really an intern, but you you're hired and you're, like, the lowest person on the totem pole. And what's different here is you have 22 year olds who are reinterviewing lifelong civil servants and are, like, in charge of them and don't have any sort of experience in the government. You know, the government hires lots of young people. It's just that they hire them at low levels, and then you work your way up.

Emanuel:

And coming in with assumptions about how things should work. Right?

Jason Koebler:

Right. Right.

Emanuel:

They're not coming in and saying, like, tell me about your job and how can we make things better. They're like, we know how to make things better. AI. How do we get there? And that Yeah.

Emanuel:

I think is is the difference.

Jason Koebler:

They're also, like, justify yourself to me, like, and if not, like, I will fire you.

Emanuel:

Or yourself, 50 year old who has given his life to the federal government. You know

Jason Koebler:

what I mean? Right. I mean, well, literally, there is reporting that's not ours, but, you know, Wired's reporting. I have no no reason to doubt it where it's like you have Elon Musk, you know, 20 year olds pressuring people in the treasury to give access to very sensitive information and and systems who have been there for a long time and have security clearances and things like that. And then when they refuse, they're they're pushed out.

Jason Koebler:

And, you know, I don't have his name in front of me right now, but one of them is like a guy who's worked at the treasury for, like, forty years or something. So not not great. And then the very last thing I'll say about Stephanie Holmes is that I believe a lot of the meetings that she's taking have to do with this fork in the road email about

Joseph:

I was gonna lie.

Jason Koebler:

Please resign if you don't wanna be here. And what's

Joseph:

the and what's the deadline for that? Because the dead we're before that deadline, and it would have happened for the next week.

Jason Koebler:

Is Thursday.

Joseph:

Right.

Jason Koebler:

And I don't think we talked about it, but the by and large, it's like Elon Musk, sent an email to everyone at the entire federal government saying, if you don't wanna work here, respond, resign to this email, and you'll get a severance more or less.

Joseph:

Which he which he did at Twitter.

Jason Koebler:

Which he did at Twitter.

Joseph:

He didn't directly send this email, but it it's like the same subject line and the same vibe. Yeah.

Jason Koebler:

Yeah. And, like, at these meetings, they just start calling it fork, which I think it I just find that to be interesting. I know someone who works in the federal government who I I've talked to about this, just like casual acquaintance. And they're they're like, everyone's just calling it fork. Fork is Thursday.

Jason Koebler:

Fork is Thursday, which is kind of interesting. And they're like, it's pretty wild how banal this has become because they're talking about potentially thousands and thousands of federal employees losing their jobs or or leaving or quitting.

Joseph:

Yeah. And the ones I've spoken to, in the government, you know, either themselves or their colleagues, they're terrified for their livelihoods, for their health insurance for them and their kids. And the deal is basically, as you say, okay, Thursday, that's the deadline. You have to decide whether to stay for this very intense season, as Shed put it from the, earlier story, which is a very funny term to me as well, like, South football is that football thing or or whatever or like a TV thing? Anyway, you either have to resign then, or or you're in.

Joseph:

And I don't think we really got into this, but in the audio that we got before as well, it was like, you know, you're gonna have to come into the office. And a lot of these software engineers and coders working for the government are remote and remote first, typically. So they might not live anywhere near, an actual office. And, like, do I wanna uproot my family and get pushed, basically, by someone trying to put AI into the government? Doesn't sound like a nice decision.

Joseph:

You know? Alright. Let's leave that there. If you're listening to the free version of the podcast, I'll now play us out. But if you are a paying four zero four Media subscriber, we're gonna talk about how AI slop is probably already in your public library.

Joseph:

You can subscribe and gain access to that content at 404media.co. We'll be right back after this. Alright. And we're back in the subscribers only section, where there's, like, no, no politics. No.

Joseph:

Hashtag, hashtag, stick to tech. Hashtag, do that. I don't know. I'm tired. The headline from you, Emmanuel, AI generated slot is already in your public library.

Joseph:

What's happening inside these libraries?

Emanuel:

So I don't know if, you take advantage of your library. But if you do, in order to get ebooks, you're probably familiar with one of two services that, serve ebooks to people who borrow books from public libraries. There is one called Overdrive, which people may know from a very popular app that it has called Libby, where you can lend, not just ebooks, but I think also movies and music. And then there is another service called Hoopla. And both of these services have gigantic catalogs of books that libraries can then opt into under different models so they can offer those books to their customers.

Emanuel:

And both of them have problems, but the one I focus on in this story is called Hoopla.

Joseph:

Before you get to that, can I ask I mean, I don't mean to put you on the spot, but I'm putting you on the spot, and maybe you don't know? But when you are hiring a book through one of these apps or something similar, how exactly is that working? Like, the library has bought a certain number of licenses or something like that? Because you open the app, you're like, what well, why is that limited stock? It's a file.

Joseph:

It could be the book.

Emanuel:

Yeah. Totally. So, I'm glad you asked that because I've been working on this story for a long time, and I complained to Jason endlessly about how hard it is. And it wasn't hard because the sourcing was hard. The sourcing was a delight because I find that librarians tend to be wonderful, generous people, and they were very smart and, forthcoming with information.

Emanuel:

The thing that was very difficult is finding out the answers for exactly what you're asking because it's a very complicated world of licenses around ebooks, and it's kind of core to this whole story. The, generally, the way that it works is that there is a license per book per lender. So, yes, it is a totally artificially limited number of ebooks. And I think the book publishers and the services that I'm gonna talk about, what they would argue is that that is just a way for them to fund book publishing, which I think is fair. You know, you look at a model like, I don't know, Spotify, where you can have millions and millions and millions of streaming of a song and the artist only ends up making a few cents on that.

Emanuel:

I'm not saying that authors are making a ton of money under this model, but there's this artificial scarcity of digital copies that's in place in order for the publishers to get paid. To follow-up on that, the the the the important distinction between Hoopla and Overdrive is that with Overdrive, they they have this giant catalog of books that they offer, and librarians can pick and choose which books they wanna offer to their customers in a way that is not that different from how they pick physical books. Right? They they look at their, customers. They look at their library, and they say, this is an important book for us to have.

Emanuel:

This is what people want. This is what people, I think, want. And you just put it on the bookshelf or you include it in the Overdrive library that your library has access to. Hoopla, on the other hand, doesn't give you many options in which books from the catalog you can offer your customers. You kind of pay for access to the catalog, and then you pay for what people lend.

Emanuel:

Right? So you pay taxes hoopla, and then when I take out Joseph Cox's, Dark Wire

Joseph:

Dark Wire, the incredible true story of the largest stingray operation ever.

Jason Koebler:

I would never take it out from the library because I've purchased several copies.

Emanuel:

Yeah. Right. And listen to the audiobook, which you can also get from the library, by the way. Yeah. Mhmm.

Emanuel:

But every time a customer takes out one of those copies, Hoopla gets paid, the publisher gets paid, Joseph probably doesn't get paid, but theoretically, he does get paid. Yeah. I I don't think so. Right. Yeah.

Emanuel:

So so they they have some ways to control what people, can take out. For example, if you have a very limited budget, which many libraries have very limited budgets, you can set a limit on, we don't want anyone to lend a book that costs more than $5 per lend. But beyond that, you don't have control over what people could take out. They just have access to the whole catalog. Now Hoopla, when it is offering its services, it celebrates the huge number of books that it has.

Emanuel:

I actually don't have the number in front of me, but, you know, many, many, many thousands of books. And that's an impressive number, and you as a librarian want people to have access to as much information as possible. The issue is that, the number of low quality books on Hoopla, has increased a lot according to the librarians I've talked to because more and more low quality AI generated books are making it into the service, which means if you go to your local library and you use their, you know, offering of Hoopla to take out a book about, let's say, Elon Musk, and you just take the first book that comes up in a search, but then you get home, you open up your ereader, and you see that it's like a low quality e generated, AI generated book. That means that your library paid with taxpayer dollars for AI generated slop, and this is a problem that librarians tell me is, like, getting pretty bad.

Joseph:

Yeah. I mean, that's that's really, really bad, especially when you bring the the financial stuff as well. Because, like, it's not just, oh, that's annoying. I got an AI book. Like, there's actually an impact on libraries as well.

Joseph:

You mentioned that Musk one. Are there any more examples? And when and when you say AI slop, I know we use that term a lot. Other people use it a lot. Like, what do you mean?

Joseph:

Just they have, like like, are they readable? Are they really bad? Are they probably made with chat GPT or something similar? Like, more specifically, what was the deal?

Emanuel:

Yeah. I mean, they're definitely readable because, LLMs are very good at producing readable text. That's what they do. But that doesn't mean that the content is of a high quality. I think one example, I was talking to a librarian, and he said something that worries him is, he told me just go in and search for fatty liver diet, which is medical information.

Emanuel:

And I got a ton of cookbooks and lifestyle books about how to eat if you have a fatty liver or a liver condition. And I just randomly clicked on one of them, and I was reading the text, and it became immediately obvious that it is low quality AI generated text, probably chat GPT, just summarized bullet points, some recipes in there. And I looked up the author who had a headshot on Amazon. It is clearly an AI generated author. The author's name is Magda Tangi, an author with zero.

Jason Koebler:

Can't believe you think Magdhatangi is not real.

Emanuel:

Dude, super fake. Check out the picture in the story. It's just like an extremely fake, woman. And

Joseph:

Can you briefly talk about how you can tell that? And I don't even mean specifically here. It's just like you, like many other people, have now developed sort of a muscle for just sort of detecting when something is fake. Is it is it is it more that, basically?

Emanuel:

It depends on the image, but I can definitely tell pretty quickly a lot of the time. Sometimes things will get past me for sure, but, there is a type of AI generated phase that I believe comes from this person isn't real. I forget the name of the website.

Jason Koebler:

This person does not exist.

Emanuel:

This person does not exist. Right. And all those AI generated faces, one quick way to spot them is the eyes and the mouth are always in the same location in the image. And then there are imperfections that you can see with AI. So I'm looking at Magda's beautiful face here, and she has you know, she looks like a, I don't know, a a a nice middle aged woman, but she has one earring on one ear, and it doesn't, like, fully connect with her ear.

Joseph:

Yeah. It's just one earring on her left ear. Yeah. You're right.

Emanuel:

And then

Joseph:

Then that's if that's her style, that's fine. Yeah. Totally. It's weird in a professional headshot, allegedly. Yeah.

Emanuel:

Yeah. And then also, you know, I I I gave the the the image to, Reality Defender, which has a software tool that goes in and kind of examines the image pixel by pixel and gives a probability that the image is, AI generated, and they said that it's 85% likely to be AI generated. Anyway, all that being said, I didn't find anything super concerning about the, the book is called the fatty liver diet cookbook, two thousand days of simple and flavorful recipes for a revitalized liver. I didn't find anything horrible in there, but it was clearly like a churned out, AI generated book. And we have seen I don't know if you remember, but I was writing about, AI generated recipes on Instacart last year, and those recipes had ingredients that didn't exist.

Emanuel:

So that kind of stuff is common. So there's that version of it. Something that is very popular is biographies, which is something I've written about before. Like, you can go to Hoopla, type in the name of most celebrities, and you'll get a bunch of very low quality, quote, unquote, biographies of those people. And I read a couple of those, and I think I didn't find horrible information, but it's trash.

Emanuel:

It's just a garbage book. And one can easily imagine how someone who legitimately wants to read a biography of Taylor Swift would end up with one of these low quality books and then read it and realize that it's trash and then would have wasted their time and would have wasted the library's money. And to be clear, librarians are telling me this is happening. Right? This isn't a theoretical problem.

Emanuel:

They're hearing from their customers that people are lending these bad books by mistake and are unhappy with them.

Joseph:

How did you first hear about this? Did a librarian reach out or something? Or

Emanuel:

A very nice librarian reached out and, told me about this. And I think something I've mentioned this several times, but I keep saying it because I wanna keep hearing from these people. The reason I am interested in this and the reason I, waited through the jargon and minutiae of libraries in order to to to do this story is that I'm very interested in the people who are on the front line of dealing with the onslaught of AI generated slop. And librarians, I think, are such a good example of that. I did something similar with Wikipedia editors.

Emanuel:

But librarians might even be a better example because it is such an old profession, and also they have a lot of experience with this problem even before AI. So we call it AI slop. I thought it was really interesting that this term has existed in the library world called vendor slurry. And that is just the idea that these services have so many vendors, meaning different publishers who are feeding Hoopla different books, that they're used to to to low quality books, and they're used to filtering them out. It's just that now there's so much more of them because they're so much easier to produce that it is becoming a bigger problem.

Joseph:

Yeah. So I haven't written this yet. So anybody listening, please don't go and steal it. I mean, it's not the biggest scoop in the world or anything. It's just we've been too busy.

Joseph:

But I showed you there's a publisher on Amazon, and they seem to just publish summaries of other people's books. And the reason I found it is because there's a summary of my book. And I bought it and downloaded it, gave them five pounds or whatever, and then reading through it. And it's just like summaries of my original reporting. My book is like 90% original reporting.

Joseph:

Spent a very, very long time, investigating it. It's just summaries of each chapter. I'm gonna probably write about that. You know, ping Amazon and be like, is this okay? Try and ping the publisher, but they have a very, very small online presence.

Joseph:

Just briefly, I showed you that I even think I showed you one screenshot, and I think Jason said this as well, but, like, it's almost certainly AI generators. Like, what's the deal there, do you think?

Emanuel:

Summaries are just probably the most common type of a a gen AI generated books in these services from what I've seen. And the same is true for AI generated books on Amazon. And I think the reason for that is that it's just a very easy way to produce a book that you could argue in front of Amazon or in front of Hoopla is a new original book. So one would take the text of your book, feed it to something like ChatGPT and say, like, summarize each of these chapters, and you just publish that. Every librarian that I talk that I talk to singled out one publisher called IRB Media that is on Hoopla that has dozens and dozens and dozens of, book summaries on Hoopla.

Emanuel:

Any book you can imagine, there's a book summary of it from IRB, BEDIA, and just a quantity of books. It doesn't make sense that any company would be able to summarize so many books so quickly. And I don't think there's anything there's nothing wrong with a book summary being offered via a library in principle, I think.

Joseph:

Yes. Like like cliff notes or something. They're like literally cliff notes. Yeah. Yeah.

Emanuel:

It's just that, again, they're low quality. So cliff notes, you can buy one of those. It will summarize it for you and maybe, like, I don't know, give you a few takeaways as a as an English teacher would about a tale of two cities or something like that. But these don't do that. These are just as you can stop from reading your own summary of your own book, it's just a summary of what is in the chapter.

Emanuel:

There's no added value to it of any kind.

Jason Koebler:

Can I say two things before we end? Sure. One, Libby gang for life. Libby is way better than hoopla. Much better books on there.

Jason Koebler:

I get most of my audio books from there. And two, we said we wanna talk about politics, but libraries are under attack, political attack and funding attacks, etcetera, etcetera. And this stretches them even further thin, which we did talk about. But one of the reasons they're stretched so thin is because their budgets have been getting cut. They've been attacked by groups like, Mobs for Liberty, etcetera, etcetera, to ban books and criminalize some of that sort of thing.

Jason Koebler:

So, I mean, you have to take it all in that context as well.

Emanuel:

Yeah. This is another thing that all the librarians I talked to really stressed is that this is something that is difficult for them to talk about because they don't want to censor people's access to information. It's not as if they don't want you to have a a biography of Elon Musk because they think he's a conservative or something. They do. They just want you to have quality information, and they're worried that any conversation about this will make it seem like they have a political agenda, which is is not the case.

Joseph:

Last thing I'll say is that you should go listen to Dakota with Neil Patel, his podcast about Big Ideas Never Problems, pass part of the Vox Vox podcast media network. Oh my god. I'm so sorry.

Jason Koebler:

Vox media podcast network.

Emanuel:

What is that?

Joseph:

Yeah. I know. So anyway, he just had one talking to the bookshop CEO, which is trying to basically be a middle ground like, an ecommerce platform for local bookshops. And they bring that up because some of the well, all of these local bookshops, they're gonna have similar motivations and that we just wanna get books into people's hands. You know what I mean?

Joseph:

And we have to deal with all this bullshit. Alright. Let's let's let's let's

Jason Koebler:

let's What about the one with James Don't, the man who saved Barnes and Noble?

Joseph:

I did listen to that as well, and I think that was good. Yeah. It's a good show.

Jason Koebler:

Oh, isn't there anything about the the recovery of Barnes and Noble? It's very fascinating.

Joseph:

Mhmm.

Jason Koebler:

Not saying Barnes and Noble is good, but very fascinating.

Joseph:

Well, I think I think they're good now. Right?

Jason Koebler:

Well, they're good now, but they're still a big company, and they're still competing with, like, local book stop short stores and stuff. So it's like, I don't know. I went into one the other day, and it was pretty cool. So that's a good

Joseph:

franchise model if I remember correctly.

Jason Koebler:

The the way that he saved the way that he saved Barnes and Noble is that he let the people who work there recommend books.

Joseph:

Right.

Jason Koebler:

That's it.

Joseph:

They've run their they've run their own store, basically. Yeah. Yeah. It's true. Alright.

Joseph:

Let's leave that there, and I will play us out. As a reminder, four zero four Media is journalist founded and supported by subscribers. If you do wish to subscribe to four zero four Media and directly support our work, please go to 404Media.co. You'll get unlimited access to our articles and an ad free version of this podcast. You also get to listen to the subscribers only section where we talk about a bonus story each week.

Joseph:

This podcast is made in partnership with Kaleidoscope. Another way to support us is by leaving a five star rating and review for the podcast. Here is one of those reviews from Cosmo Grafia. I've been binging all the episodes after discovering this podcast. The journalism is fantastic.

Joseph:

It's thorough and in-depth, and there's a lot the four zero four media team are reporting I haven't heard about anywhere else. Thank you so much. We'll see you again next week.