DOGE’s $1 Federal Spending Limit Is Straight Out of the Twitter Playbook

WIRED’s director of business and industry, Zoë Schiffer, and Katie Drummond, global editorial director, talk about credit card freezes and AI technology at DOGE, and how each is a move from the Twitter playbook.

Articles mentioned in this episode:

• DOGE Puts $1 Spending Limit on Government Employee Credit Cards
• Elon Musk’s $1 Spending Limit Is Paralyzing Federal Agencies
• DOGE Is Working on Software That Automates the Firing of Government Workers
• Elon Musk’s DOGE Is Working on a Custom Chatbot Called GSAi
• Trump’s FDA Cuts Are Putting Drug Development at Risk

You can follow Katie Drummond on Bluesky at @katie-drummond and Zoë Schiffer on Bluesky at @zoeschiffer. Write to us at uncannyvalley@wired.com.

How to Listen

You can always listen to this week’s podcast through the audio player on this page, but if you want to subscribe for free to get every episode, here’s how:

If you’re on an iPhone or iPad, open the app called Podcasts, or just tap this link. You can also download an app like Overcast or Pocket Casts and search for “uncanny valley.” We’re on Spotify too.

Transcript

Note: This is an automated transcript, which may contain errors.

Katie Drummond: Welcome to WIRED’s Uncanny Valley. I’m WIRED’s global editorial director, Katie Drummond. Today on the show: credit card freezes and how DOGE is using AI. I’m joined today by WIRED’s director of business and industry, Zoë Schiffer. Zoë, welcome to Uncanny Valley.

Zoë Schiffer: Thank you so much, Katie.

Katie Drummond: And we obviously know you well on this show, because you cohost our Thursday episodes with Mike and Lauren.

Zoë Schiffer: Exactly, yes. I’m switching sides this week.

Katie Drummond: And let’s get right into it. So Zoë, two weeks ago on February 20th, you published a story on WIRED.com about a $1 spending limit being placed on government employee credit cards. Walk us through that first story. You’ve subsequently published more reporting on that topic this week, but tell us sort of where this came from at the outset.

Zoë Schiffer: Yeah, OK. So like you said, on February 20th, employees at the General Services Administration received this memo abruptly telling them that most of the credit cards used by their office as well as every federal employee across the government, were going to have a $1 spending limit. There’s basically two types of credit cards that are used by most federal employees. There’s travel cards, which are used for all sorts of work travel, and then there’s purchase cards which are used for everything else. So think basic supplies, trainings, software licenses, all of that sort of thing. So this was almost immediately going to have a pretty severe impact on the ability for a lot of these people to do their jobs.

Katie Drummond: And what was the premise by which DOGE mandated this credit card freeze? I mean, they are sort of ostensibly, in theory, according to their mandate, trying to reduce government waste, excess spending, fraud. Was it all of that? Was the thesis of this credit card freeze was that they would root out fraudulent spending?

Zoë Schiffer: That was the inference. They actually said, I think the word they used was that they were trying to simplify the credit card program. And then they hinted that there was a lot of wasteful spending on these cards. In fact, there was a study that was commissioned, I believe in 2002, that found that by bypassing the typical procurement process needed to get goods and services at the federal level, the federal government was actually saving $1.2 billion. Katie, does the term zero-based budgeting mean anything to you?

Katie Drummond: It means less to me, Zoë, than I think it does to you. So why don’t you walk us all through what that actually means.

Zoë Schiffer: OK. Yeah. There’s basically a punch list that I have at this point of all of the things that happened when Elon Musk bought Twitter, and I’m kind of going one by one and saying, “Oh, OK. Yeah, this is happening at the federal level. Oh, this is too.” So zero based budgeting is this idea that you take a budget, you slash it down to zero, and then you force the people under you to justify every single expense. At Twitter, what this looked like is that people were kept in a conference room on Saturdays for 12 hours at a time, and they were arguing directly to Elon Musk about why a critical security software was needed at the company. And if he didn’t agree with them, he would often fire them on the spot. At the federal level, what it means is that DOGE is trying to again slash the budget down to zero and then basically see who screams. The way that they talk about this is that often the biggest grifters are the one who scream first. And so you shut off all payments and then you see what’s breaking.

Katie Drummond: First of all, what a fascinating way to run not only a company, but the entire federal government. Second, the screaming has started. So since you published that story two weeks ago, that spending limit has rolled out across federal agencies across the government. And last night, Monday night, March 3rd, you and Emily Mullin, a reporter at WIRED, published another story about how this spending limit is essentially paralyzing federal agencies. And you have a ton of specific instances in here across federal agencies documenting what the impact of this spending freeze has been. Can you walk us through a little bit more about how this is disrupting the federal apparatus?

Zoë Schiffer: Yeah, for sure. There’s basically two big buckets that I’m putting the disruptions into. There’s kind of the chaos and confusion bucket, which is that employees all just receive this memo saying you’re not going to have access to money anymore. And it’s not clear whether or not if you put expenses on your personal card, you’ll ever get reimbursed. And so suddenly, people who work at the Federal Aviation Administration and have to travel to airports around the country to test out security and safety kind of software aren’t sure, like, wait, that’s literally my job, but am I allowed to make that trip? If I do, do I put it on my personal credit card? So it’s just stalling work that was previously done with a lot of ease. And then there’s the real tangible impacts, which are a researcher at the National Institutes of Health who tests new vaccines and treatments in rodents, says he’s had to put experiments on hold because his lab isn’t able to get antibodies, which are critical to do this sort of research.

Or I talked to employees at the National Park Service who said they were literally stockpiling toilet paper because they weren’t sure that they would have access to funds. And this is obviously critical infrastructure for federal lands, for public parks. Similarly, like NPS, the National Park Service said we put a lot of our expenses like internet and cell service on credit cards. And so if those get shut off and there’s a bathroom that needs to be fixed at a national monument, suddenly we’re not going to be able to put in the work order. Work will just grind to a halt. But we had tons and tons of these examples. Employees at the National Oceanic and Atmospheric Administration said, scientists aren’t able to buy equipment used to repair ships and radars. Employees at the FDA said labs are experiencing delays in ordering basic supplies. So really what it looked like is that many people are already unable to carry out the very basic functions of their jobs. And this again, is all in the name of efficiency, but the people we talk to are saying it seems like our lives have become much, much less efficient.

Katie Drummond: And you wrote an excellent book about Elon Musk’s takeover of Twitter, now X. So you have sort of this really interesting point of view on him and his playbook. Do you get a sense that Musk and DOGE leadership that the president know about just how sweeping these credit card freezes have been? Where do we sort of position federal leadership in all of this?

Zoë Schiffer: Yeah, I think it’s a really good question. I think when we talk about Elon Musk and people in DOGE, and this could apply to Trump too, it’s really important to keep in mind that they absolutely conceptualize themselves as the good guys. They don’t see themselves as coming in and making people’s lives worse or more complicated. And so while it’s clear that they’re hearing from people who are unhappy with what they’re doing, I think that they’re hearing more about the good that their changes are doing at the federal level.

We know that people in DOGE are highlighting examples of, look at all of the fraud we’ve found. Look at all of the way waste we’ve eliminated to Elon Musk. And so I think that the credit card change, like a lot of the changes DOGE has made is really being seen as like, wow, the way that government functioned before was broken, it was wasteful, it was inefficient, and we are coming in and we’re doing all this good. And yeah, it might be annoying for people along the way, but the greater good is really what in their minds, they’re kind of keeping as the north star.

Katie Drummond: Well, good for them. Where do federal workers go from here? Is there any hope of them being able to resume some version of business as usual, some version of using credit cards or some other process to obtain, whether it’s the travel they need, the materials they need to do their jobs?

Zoë Schiffer: I think it’s really unknown. And honestly, what we heard from a USDA official for example, is the longer this goes on, the more the systems are going to break. Some agencies did have a little bit of warning that this change was coming down, and so people literally did go out and labs were stockpiling reagents and workers at the National Park Service were stockpiling toilet paper like we said. So I actually think that right now the change might not be felt as acutely as it will be in about two weeks when a lot of those supplies run out.

Katie Drummond: We’re going to take a short break. We’ll be back with Zoë Schiffer in a minute.

Welcome back to Uncanny Valley. So Zoë, I want to talk to you about how DOGE appears to be using AI in one instance, editing proprietary government software that could actually fire government workers on mass. The notion that AI could actually be used to conduct mass layoffs. As it turns out, and this was surprising to me, this is actually not a new concept for the federal government. Talk us through that.

Zoë Schiffer: Yeah, so this software, what we’re talking about right now is a program called AutoRIF that was developed by the defense department like decades ago. And this story is by our wonderful reporter, Makena Kelly, and what she found was that DOGE is taking that old software and kind of repurposing it to maybe conduct mass firings of federal workers. AI is really at the heart of the DOGE agenda in a lot of ways. And it makes sense on a philosophical level because if your whole stance in coming into government is the way things used to run is inefficient, backwards, doesn’t make sense, we’re going to come in, we’re the technologists, we’re going to make everything run really, really smoothly, then it makes sense to use AI for that purpose. But it also makes sense on a practical level because if you’re going to mass fire people and you’ve already laid off the people who would conduct those firings, then you do need to automate parts of that process.

You need to offload work that was previously done by humans and give it to machines or large language models in this case. So it looks like a former Tesla engineer appears to be overseeing AutoRIF’s development based on Makena’s reporting again. And his involvement really shows how deeply embedded Elon Musk is in every part of this process. Even ones where his fingerprints aren’t as clear, his people are the ones developing these tools and rolling out these new programs. It also touches on this story that WIRED has reported on pretty extensively, which is employees were asked to submit five bullet points detailing their accomplishments from the previous week.

Katie Drummond: There’s been so much chaos around these bullet points in these emails, and I think we’re seeing reporting indicate that those bullet points would potentially be fed into an AI to determine whether or not someone should keep their job. Is that correct?

Zoë Schiffer: I think the way to think about this is that at the most basic level, it is a loyalty test. And Elon Musk is constantly conducting these sorts of tests on the workforce. He did this again at Twitter. It was like, let’s ask employees to do something that is both simple. Just respond to this email. Tell me what you did last week. And also kind of offensive or paternalistic to employees based on what they’re telling us. But the kind of goal of it is to see who complies. And right away if you have people who don’t respond to the email, you can bucket those people in a category of maybe they’re not loyal to the new regime. Down the road, employees are already being asked to do this at some agencies every single week. And so then you can see how it would become more of a question of who is productive? Who is working on things that we find to be important? And again, who isn’t, and could we lay off in a next reduction in force?

Katie Drummond: Absolutely surreal, I will say. Now DOGE is not only editing existing government software like AutoRIF, DOGE is also exploring custom chatbots, for example, sort of developing its own AI to use within the federal government. That’s a story that WIRED broke a few weeks ago, but tell us what we know about that, about sort of the idea of developing new AI to be used across the federal government.

Zoë Schiffer: Yeah, so what we’re hearing from federal employees is that especially employees who worked at the United States Digital Service, this part of the government that was kind of repurposed to become DOGE and GSA, the General Services Administration. So again, kind of the technology arm of the government is that right away after Trump’s inauguration, they start hearing from DOGE all the time, and a lot of the requests are, can you do this with AI? Can you slap AI on this? Can you upgrade how you’re working with machine learning and large language models? So it felt kind of like this onslaught of requests about how much can we embed AI into the work that we’re doing. The line that you hear a lot when you’re talking to these people is a lot of these projects seem like they would actually take years, but DOGE thinks in days and weeks.

And so the chatbot interestingly is kind of this … It’s not that difficult to spin up a chatbot. And so I think in some ways it was kind like it made sense because a chatbot is an easy way for workers who might not be as familiar with this technology to interact with a large language model. And so if you deploy it across the federal government, maybe it can be a new search tool that employees use. Maybe it can help them boost their day-to-day productivity, but also it’s a way for the employees who are having to build these AI products to say, OK, DOGE, we’ll do this in the next few weeks. We’ll do this really, really fast. At the same time, we’re going to kind of keep an eye on the longer-term projects that they clearly want, which seems to be about how do we kind of process government data and automate the processing of that data with large language models.

Katie Drummond: And now, just to be sort of clear and generous in spirit for a minute, if I may, the federal government is a massive set of agencies. It is the largest employer as of now in the country. There’s a lot of bureaucracy, there’s a lot of data. There is sort of a lot flowing through that infrastructure. There are certainly valid and credible uses of AI within federal agencies. I don’t think anybody would argue with that. I don’t think we at WIRED would argue with that. But what potentially goes wrong? What concerns you as you sort of see the editing of existing government AI or the development of new AI projects, sort of this idea of spinning something up in days and weeks? Where should we be concerned here?

Zoë Schiffer: I think it is worth saying that if DOGE had come in and worked with government employees, so long-standing civil servants to roll out some of these projects, I think a lot of people would’ve seen this as a positive. Government is inefficient. It can be quite wasteful, and a lot of work can be successfully automated with large language models with AI. But that’s not been their stance coming in. They’ve come in with a fair amount, I think it’s OK to say with hostility, with mistrust to the people who ran the government previously. And so they’re kind of pushing those people aside.

They’re coming in at times without a lot of knowledge about how the system functioned previously, what the quirks were of this software, and then they’re kind of rolling out their projects. And the issue is that AI already makes mistakes. There can already be biases baked in. And so I think you need to do this really methodically with a willingness to roll it back if it’s not working with an eye toward what mistakes are being made, what’s being lost. And it just seems like DOGE is working quite fast and according to some people, a bit carelessly.

Katie Drummond: Right. Move fast and break things as we’ve been saying a lot at WIRED in the last few months. We’re going to take a short break, when we come back, what you need to read on WIRED today.

Welcome back to Uncanny Valley. I’m Katie Drummond, WIRED’s global editorial director. I’m joined by WIRED’s director of business and industry, Zoë Schiffer. Now Zoë, before I let you go, tell our listeners what they absolutely must read, must read on WIRED.com today, other than the stories we talked about in this episode.

Zoë Schiffer: OK. I wish I had a nice, joyful, uplifting story to talk to you about, but I have another doom and gloom story, and it’s by—

Katie Drummond: Aw-shucks.

Zoë Schiffer: I know. It’s by Caroline Haskins, who is a freelancer for us, and actually we just announced she’s joining the business desk. So exciting. She’s incredible. She’s so good. I’m so excited. And she wrote a piece that we published yesterday about how Trump and Elon Musk’s cuts at the FDA, so another administration that has experienced severe budget and staffing cuts is already putting drug development at risk. And she got this from dozens of SEC filings from pharmaceutical companies.

Katie Drummond: So between those SEC filings and what you and Emily reported yesterday about these credit card freezes, it certainly seems like we are seeing federal agencies ground to a halt here in some really consequential ways.

Zoë Schiffer: Yeah. I mean, it’s interesting because the drug companies, the pharmaceutical companies aren’t even saying, “The FDA isn’t approving our drugs, and so these drugs can’t come to market.” They’re saying this agency was already so slow moving by design because the stakes are very, very high when you’re talking about drugs and medicines. And so staffing cuts, budget cuts. The worry is that this will grind to a halt. And if you’re a pharmaceutical company that’s deciding between continuing to produce a drug that’s already been approved or putting a lot of time, energy, and resources, money behind the development of a new drug that you’re not sure will get FDA approval, suddenly you’re going to see less of that and more of the kind of, OK, we’ll just pour money into the existing product pipeline. And that has really serious implications for people who might need these new therapies.

Katie Drummond: Zoë, thank you for all of the joy that you brought to our show today. Thank you for joining me. Genuinely though, fascinating stuff and so grateful for your reporting and the team’s reporting.

Zoë Schiffer: Thank you so much for having me.

Katie Drummond: That’s our show for today. We’ll link out to all the stories we talked about today in the show notes. Make sure to check out Thursday’s episode of Uncanny Valley, which is all about Silicon Valley’s pro-natalist movement. If you like what you heard today, make sure to follow our show and rate it on your podcast app of choice. If you’d like to get in touch with any of us for questions, comments, or show suggestions, write to us at uncannyvalley@wired.com.

Amar Lal at Macro Sound mixed this episode. Jake Lummus is our studio engineer. Jordan Bell is our executive producer, Condé Nast’s head of global audio is Chris Bannon. And I’m Katie Drummond, WIRED’s global editorial director.

Goodbye.