Well, nothing’s going to slow down for the foreseeable future, is it. Alas. Before we get into Elon Musk’s quest to automate the government, a quick programming note: This newsletter is made possible by readers who chip in a few bucks a month to make it possible (thank YOU). We are quite clearly moving into dark and uncertain times, ruled over by tech oligarchs and their dreams of AGI—I plan on using this space to investigate, document, and break down what they’re doing. I can only do that with the support of readers like you. If you find value in this writing and analysis, and are able, consider becoming a paid supporter, so I can keep at it. And share this with other interested humans, too—it’s all much appreciated. And with that, onwards!
Over the last week or so, a freshly empowered Elon Musk and his now-infamous intern squad at DOGE led a breakneck incursion into a number of federal agencies—the foreign aid-distributing USAID, the federal infrastructure-managing General Services Agency, and, perhaps most alarmingly, to the Department of the Treasury, where Musk and co now have direct access to the nation’s payment system and to the private information of millions of Americans.
Some key developments among the blitz:
-8,000 government websites were taken offline, many from the CDC and the Census Bureaus. “The purges have removed information about vaccines, veterans’ care, hate crimes and scientific research, among many other topics,” the New York Times reports.
-Millions of federal workers have received “fork in the road” emails, offering them 8 months of severance if they resign by Thursday, February 6th.
-An automated system is now scrubbing pronouns from government email signatures.
-In a Monday morning meeting, ex-Tesla engineer Thomas Shedd, the newly appointed director of the GSA’s Technology Transformation Services, told the department that it would be pursuing an “AI-first strategy.”
Got all that? What’s happening at federal government agencies is, among other things, a stark portrait of why power seeks to automate—and why, in enterprise AI, it has found an ideal vessel for corporate and administrative automation.
It’s not just about displacing labor, although that is of course a key thrust—in this case DOGE is seeking to winnow the federal workforce by driving out nonpartisan workers with the lowest tolerance for the new, ideologically motivated regime, and promising to replace the bulk of their tasks with AI. The Musk lieutenant Shedd emphasized that “AI coding agents” would be prioritized and made available for all agencies. WIRED reports that “Shedd made it clear that he believes much of the work at TTS and the broader government, particularly around finance tasks, could be automated.”

It’s also, and perhaps above all, about control. Automation necessitates the narrowing of scope, of information input into a system, of possibility—so that a job or a task or work can be more predictably and repetitively performed on behalf of an administrator. In DOGE we see the logic of automation—of enterprise AI—being imposed by a nascent oligarchic state. We thus see less information available to the world, fewer options available to the humans working to provide it, fewer humans, period, to contest those in power, as that power concentrates in their hands.
This, in short, is why Elon Musk is trying to automate the federal government.
There was a somewhat lazy joke I took to making in the purgatorial period between Trump’s election in November and his taking office in January, when few yet knew how seriously to take DOGE or Elon Musk’s role in the forthcoming administration. With Musk’s incessant proclamations about cutting trillions from the federal government on X and in the Wall Street Journal’s op-ed pages, and his obsession with AI, it seemed to me that a core assumption undergirding the moves was that AI might be made to perform federal workers’ jobs—and hey, he owns an AI company, and knows a thing or two about securing big federal contracts. We were about to see the rise of government by Grok, was the joke.
Ha, ha.
The problem about making jokes like that at a time like this is that they are prone to come true rather quickly and become unfunny even faster. It doesn’t even matter if the notion itself—that millions of government workers can be replaced by AI systems that have trouble counting to 50, are still prone to hallucinations, cannot ably function in physical spaces, and have no institutional knowledge—is absurd on its face.
Per WIRED:
"This does raise red flags,” a cybersecurity expert … told WIRED on Monday, who noted that automating the government isn’t the same as automating other things, like self-driving cars. “People, especially people who aren’t experts in the subject domain, coming into projects often think ‘this is dumb’ and then find out how hard the thing really is.”
Or if the idea that anyone should treat the agency that manages the $500 billion in federal buildings and property as “a software startup” is functionally meaningless. WIRED again:
Shedd instructed employees to think of TTS as a software startup that had become financially unstable. He suggested that the federal government needs a centralized data repository, and that he was actively working with others on a strategy to create one, although it wasn’t clear where this repository would be based or if these projects would comply with privacy laws. Shedd referred to these concerns as a “roadblock” and said that the agency should still push forward to see what was possible.
These notions—AI can replace workers, the government should function like a startup—are not meant to describe reality; they are meant to create a permission structure for those in power to obtain more of it. Here, AI will either allow Trump and Musk to install more loyalists, hollow out the administrative state, or degrade the quality of services once provided; all outcomes that favor Trumpism, and, I guess, Muskism. The startup mentality, meanwhile, seeks to give license to break laws, in the name of progress, of disruption, of building the future.
Same as it ever was: Way back in the early days of the Industrial Revolution, early factory owners deployed automation to deskill workers, to justify employing precarious and child laborers, and as a means of circumventing long-held laws—all to produce more products at lower quality, and to concentrate profits, and power, in fewer hands.
Note that while many AI companies and executives try to pitch automation as making people’s lives easier, DOGE cuts right to the heart of it. This isn’t about making anyone’s working life better—in fact, it’s actually going to mean more work, for you, the worker.
Here’s Shedd again, this time from recorded audio of the GSA meeting obtained by 404 Media:
“as we decrease the overall size of the federal government, as you all know, there's still a ton of programs that need to exist, which is this huge opportunity for technology and automation to come in full force, which is why you all are so key and critical to this next phase,” he said. “It is the time to build because, as I was saying, the demand for technical services is going to go through the roof.”
“Which means things are going to get intense. Like across the board in every agency, the demand on all of us is going to go up,” he added. One employee asked if it is “currently illegal to work more than 40 hours a week. Is that going to change?”
“Unclear at this point,” Shedd said
Many have pointed out, quite accurately, that Musk is poised to do to the federal government what he did to Twitter—to strip it to the bone, hollowing it out so those who remain do so either out of necessity or ideological alignment, and not caring much when it breaks down. It’s going to be haphazard, and dumb, and maybe evne funny in a gallow’s humor kind of way, because it’s the actual federal government and not a Web 2.0 app. But we should be paying attention to where he builds on that metaphor, too—where Musk and his lackeys are using the logic of AI and of technological disruption to justify his further integration with, and control over, crucial parts of the state.
This is government by Grok, so to speak, and stupid as it all is, Musk is now in an unprecedented position to extract data, labor, contracts, and assume powers no unelected man has ever held over the nation. And that is why he automates.
A quick final note: You may have noticed I linked an awful lot to WIRED and 404 Media in this piece, and that is because both of those outlets have quickly proven to be indispensable in fearlessly reporting on the Trump-Musk era. I heartily recommend subscribing to each, if you can. All the best out there, and hammers up.
That's the trouble with Didn't Earn Its..... they have no idea.
Trump and Musk have never done the work...
Thanks for the though-provoking piece.
Government by Grok echoes Harari's warning that AI-driven automation risks concentrating power in the hands of elites. This will obsviously enable unprecedented surveillance and control and in so doing undermine democratic institutions.
Musk’s drive for factory-like efficiency and automation as applied to governance means stripping bureaucracy down to its algorithmic core while consolidating authority under tech oligarchs. On the one hand, this move will sure create efficiency gains, cost savings, and in many public facing sectors, improved service (!), but the coin flips between techno-optimism and techno-fascism. What fuels Musk’s talent for automation also enables AI-powered political control. Here efficiency is an attractive mask hiding authoritarian rule.
Parallels to the CCP’s surveillance state are hard to ignore. Both the US's emerging regime and the CCP increasingly rely on AI to tighten control, suppress dissent, and optimize governance through mass data collection and predictive enforcement.
Keep up the insightful and pressing reporting and commentary!