Since I live in Minneapolis, I'd say we understand AI's risks to democracy better than Dario, given that AI has already been used to identify and map not only ICE targets but also to track protestors and journalists within the US. The pro-AI internet writers seemed to miss when this was actively happening earlier this year, and I'm not sure why because it made the national news. Since I'm old enough to remember decades of discussions about the PATRIOT act, the use of Palantir's AI-enabled tools to track citizens has been as shocking as any other revelation from Operation Metro Surge.
The logic that anything Hegseth does is a media event is a bit daft to me, since he is not currently a media professional. He is actively hurting people in a multifront war.
Like many other Trump officials, he is extremely familiar with media cycles, but nothing he is doing is unique. Every senior official in the administration is deeply savvy about the current state of how tech and media interact to keep them in power. I mean, they built their own arsenal of influencers, media companies and social networks to promote their own authoritarian actions and narratives. They are not doing media; they are doing textbook 20th century totalitarianism applied to 21st century media systems.
In no way are you like Pete Hegseth, Jasmine. You downplay what is sinister and violent in hopes it will turn out ok, which is very sweet but also not particularly grounded in the reality of the current moment.
Regardless, I'd encourage workers in AI not to just think about elections, but to engage in loud and active conversation-- and maybe even indulge in some activism-- regarding how the United States is currently using AI tech in all the worst ways. They are taking the paycheck, so they have a responsibility to chew on this morality and use it to guide their actions as adults all the time, not just every couple of years in November.
Media, propaganda, and aesthetics — creating content about the violence in addition to doing the violence — has always been a core part of the fascist’s playbook. I don’t point to the spectacle in order to downplay what’s happening. I think Minneapolis was Miller’s ICE making a spectacle of brutalizing immigrants and protesters in one city the same way that Hegseth wants to make a public example of Anthropic. I think the admin wants to do media precisely because they want to create a chilling effect on speech and dissent.
If you read this essay as me trying to downplay the admin, rather than trying to encourage the tech community to take seriously its authoritarian actions, I can’t really persuade you otherwise.
Imagine a world in which all the tech companies who work with the DoW said "we stand with Anthropic" and stopped working with the government in response to this tantrum. That would be glorious indeed.
Great article with a more political bent than usual. It was especially interesting seeing how you square the AI company and Trump world circle. As I was reading the news I had the opposite reaction to you. I thought this was fairly typical in political terms, a cabinet member thumbing his chest demanding obedience (as with tariffs or with immigration) before losing interest or another story getting more prominence. I thought big tech companies knew or were more aware of the political risks associated with this Admin. So I was surprised by the significance of the story, just as you were surprised that Big Tech thought its support bought it protection from the political risk. So I was on the opposite side of your trade.
The fascist argument is interesting, I agree in sentiment, but I think a more persistent feature is the drumming uncertainty that comes with swinging actions. Whereas past admins tried to be consistent and not coerce the fact this so persistent means AI safety à la Biden would be impractical. You once wrote/said about the Tech shift to the right that the idea that the Gov could restrict startups was part of the explanation (Marc Andreesen being upset about unbanking) in some of the tweets about how one shouldn't fund AI startups I see just that. After all for a large part of the economy there are already new arbitrary rules that make businesses harder. Tech seems now more caught up and lobbying for the vibes has its limits.
Overall the piece felt more direct and less prosaic with a nice call to action. More Kevin Roose less Susan Sontag. Wonderful in any case!
Great piece. Makes me wonder why any AI company would choose to build in America any longer? If this is how the government reacts over reasonable requests from a safety-first AI firm, who as you noted already had these terms and conditions baked into existing government contracts, why should any other firm trust the feds?
I fear this will have a massive chilling effect on US tech and business generally. It’s crazy to think that in 2026 there may be better rule of law and business certainty in a jurisdiction like the UAE than in America.
I think you nailed it on the intent of this action - it’s all about intimidation. And I’ve also written about this new legal and economic approach by the current admin, although I’ve called it “American State Capitalism.”
As one of my political science profs said back in the day, “When the rules of the game no longer matter, everyone loses the game.”
This is amazing (👏, Jasmine!) but I had one question: no one forced Google or Apple to make the change in their maps from “Gulf of Mexico” to “Gulf of America,” right? I thought this was something they chose, thinking that placating the WH was easier than taking a stand.
Jasmine I have been waiting so eagerly for this post
Since I live in Minneapolis, I'd say we understand AI's risks to democracy better than Dario, given that AI has already been used to identify and map not only ICE targets but also to track protestors and journalists within the US. The pro-AI internet writers seemed to miss when this was actively happening earlier this year, and I'm not sure why because it made the national news. Since I'm old enough to remember decades of discussions about the PATRIOT act, the use of Palantir's AI-enabled tools to track citizens has been as shocking as any other revelation from Operation Metro Surge.
The logic that anything Hegseth does is a media event is a bit daft to me, since he is not currently a media professional. He is actively hurting people in a multifront war.
Like many other Trump officials, he is extremely familiar with media cycles, but nothing he is doing is unique. Every senior official in the administration is deeply savvy about the current state of how tech and media interact to keep them in power. I mean, they built their own arsenal of influencers, media companies and social networks to promote their own authoritarian actions and narratives. They are not doing media; they are doing textbook 20th century totalitarianism applied to 21st century media systems.
In no way are you like Pete Hegseth, Jasmine. You downplay what is sinister and violent in hopes it will turn out ok, which is very sweet but also not particularly grounded in the reality of the current moment.
Regardless, I'd encourage workers in AI not to just think about elections, but to engage in loud and active conversation-- and maybe even indulge in some activism-- regarding how the United States is currently using AI tech in all the worst ways. They are taking the paycheck, so they have a responsibility to chew on this morality and use it to guide their actions as adults all the time, not just every couple of years in November.
Media, propaganda, and aesthetics — creating content about the violence in addition to doing the violence — has always been a core part of the fascist’s playbook. I don’t point to the spectacle in order to downplay what’s happening. I think Minneapolis was Miller’s ICE making a spectacle of brutalizing immigrants and protesters in one city the same way that Hegseth wants to make a public example of Anthropic. I think the admin wants to do media precisely because they want to create a chilling effect on speech and dissent.
If you read this essay as me trying to downplay the admin, rather than trying to encourage the tech community to take seriously its authoritarian actions, I can’t really persuade you otherwise.
Imagine a world in which all the tech companies who work with the DoW said "we stand with Anthropic" and stopped working with the government in response to this tantrum. That would be glorious indeed.
Great article with a more political bent than usual. It was especially interesting seeing how you square the AI company and Trump world circle. As I was reading the news I had the opposite reaction to you. I thought this was fairly typical in political terms, a cabinet member thumbing his chest demanding obedience (as with tariffs or with immigration) before losing interest or another story getting more prominence. I thought big tech companies knew or were more aware of the political risks associated with this Admin. So I was surprised by the significance of the story, just as you were surprised that Big Tech thought its support bought it protection from the political risk. So I was on the opposite side of your trade.
The fascist argument is interesting, I agree in sentiment, but I think a more persistent feature is the drumming uncertainty that comes with swinging actions. Whereas past admins tried to be consistent and not coerce the fact this so persistent means AI safety à la Biden would be impractical. You once wrote/said about the Tech shift to the right that the idea that the Gov could restrict startups was part of the explanation (Marc Andreesen being upset about unbanking) in some of the tweets about how one shouldn't fund AI startups I see just that. After all for a large part of the economy there are already new arbitrary rules that make businesses harder. Tech seems now more caught up and lobbying for the vibes has its limits.
Overall the piece felt more direct and less prosaic with a nice call to action. More Kevin Roose less Susan Sontag. Wonderful in any case!
Great piece. Makes me wonder why any AI company would choose to build in America any longer? If this is how the government reacts over reasonable requests from a safety-first AI firm, who as you noted already had these terms and conditions baked into existing government contracts, why should any other firm trust the feds?
I fear this will have a massive chilling effect on US tech and business generally. It’s crazy to think that in 2026 there may be better rule of law and business certainty in a jurisdiction like the UAE than in America.
I think you nailed it on the intent of this action - it’s all about intimidation. And I’ve also written about this new legal and economic approach by the current admin, although I’ve called it “American State Capitalism.”
As one of my political science profs said back in the day, “When the rules of the game no longer matter, everyone loses the game.”
guess the AI populists were right!
This is amazing (👏, Jasmine!) but I had one question: no one forced Google or Apple to make the change in their maps from “Gulf of Mexico” to “Gulf of America,” right? I thought this was something they chose, thinking that placating the WH was easier than taking a stand.