When SpaceX filed for its IPO, we went through the S-1 to point out just how far off the mark Elon Musk’s predictions about his Twitter takeover turned out to be. Indeed, the failure of Twitter as a business was so catastrophic, Musk had to hide it by merging all his Xs together: he had xAI buy X (formerly Twitter), and then had SpaceX buy xAI. But when Musk was buying Twitter and needed investors, he made bold promises about how he would succeed — promises the media and his investors treated as perfectly reasonable. He said he would take Twitter’s ad revenue from $4.5 billion in 2021 to $12 billion in 2027 (and then also many more billions in subscription revenue).
Now that SpaceX is a public company, he has to report some of that revenue, though it’s all merged and buried in footnotes. But there’s enough info in there to show that advertising revenue is way, way below what it was when he took over the site. And it’s shrinking.
In the 10-Q, all of X’s revenue is now (misleadingly) labeled “AI” revenue, but even then the reporting has to admit ad revenue took a big hit, dropping $160 million compared to the first half of 2025 (which was already way, way, way lower than when Elon took over the site), going from $870 million in 2025 to $710 million in the first half of 2026. The company tries to claim that this decrease in ad revenue is because they “transitioned to a new advertising platform which impacted ad sales for a short period of time.” Which is a curious explanation, given that on the analyst call, CFO Bret Johnsen credited that very same “overhauled” ad-tech platform for the very slight quarter-over-quarter uptick in ad revenue. The same overhaul apparently both tanked the business and rescued it, depending on which direction the number was moving.
And even the claim that the overhaul increased ad revenue is suspicious given that if you compare the 2nd quarter from last year to this year… ad revenue dropped from $426 million to $367 million, year over year. Some overhaul.
I kinda think that if your transition to a new ad platform decreases ad sales by $160 million, someone fucked up badly somewhere. Especially when it’s a nearly 20% year-over-year drop.
Let’s break all this down: In the second quarter of 2022, the last full quarter before Elon took over, Twitter brought in $1.08 billion in ad revenue. In the second quarter of 2026, X brought in $367 million. That’s $713 million a quarter, gone. Elon has wiped out two-thirds of Twitter’s ad business over the exact period in which he promised investors he’d nearly triple it.
And it’s not for lack of trying. Over the past year X has made a ton of experiments and revamps to improve the experience (though the main person in charge of all that, just left). I’d argue that the company was just rebuilding all of the stuff Elon killed after first taking over the company, but none of it has brought advertisers back.
Still, it seems like someone should be pointing out how far off the mark Elon has been. He promised twelve billion dollars in ad revenue within the next year. That’s three billion a quarter. X just did $367 million — about an eighth of the pace, and shrinking. You can talk all you want about his rockets and robots, he’s not magically going to grow ad revenue 700% in the next 18 months.
Last week we wrote about how Elon Musk’s xAI had filed a lawsuit to attempt to block Minnesota’s anti-nudify app law. As we tried to explain, even if you (reasonably, understandably) dislike both Elon and “nudify” apps, there were real problems with the Minnesota law. In particular, it was not narrowly tailored to just target truly harmful image edits. Indeed, it wasn’t even limited to the non-consensual use of the tech. The state’s Supreme Court had already handed the Minnesota legislature a clear roadmap for drafting a law like this that would pass strict scrutiny. The legislature ignored it.
However, the lawsuit was filed just days before the law was set to go into effect, which was called out by Minnesota’s Attorney General in arguing against a temporary restraining order:
First, X.AI’s lack of diligence confirms that a TRO is unnecessary. The company waited until the last minute to sue…
And that seemed to influence Judge Donovan Frank, who denied the motion for the TRO mainly because Elon waited until the last minute to file.
The Court respectfully denies the request for a temporary restraining order before tomorrow. xAI filed the motion on July 29, 2026, nearly three months after the law was signed, and only three days before the law is set to take effect. Such a delay in bringing the action and the motion suggests that harm is not immediate.
And, sure, it was kind of silly for Musk to wait right up until the law was set to go into effect, but that’s not all that rare with challenges to these kinds of laws. I also find the court’s suggestion that the harm isn’t immediate a bit odd, given that (as xAI had rightly pointed out) the potential liability under this law is massive: $500,000 per “access, download, or use.” That means any single use of Grok to edit an image that violates this law (which, as we discussed, goes way beyond nonconsensual sexual imagery, and could even cover someone editing a photo of themselves in a way they endorse) could lead to a huge bill for the company. And it’s now in effect, meaning in theory Minnesota’s AG, Keith Ellison, could already seek fines against the company — though there’s no indication that his office has done so yet. And while it may be politically appealing to try to enforce immediately, that may play badly before the court when there are hearings coming up in a few weeks on a preliminary injunction.
It’s also unclear if xAI actually changed anything on its end. In its filing, the company said that if the law went into effect, it would need to restrict access to certain features:
Confronted with $500,000-per-image strict liability and no safe harbor, xAI has no practical choice but to restrict Grok Imagine’s image-editing features in various ways when the statute takes effect on August 1, 2026
But the law has gone into effect, and as far as I can tell, there’s been no announcement of any changes. It’s possible such changes have been made already and just not announced. But it does come off as a bit weak to file a lawsuit on Monday saying that “if this law isn’t fixed by Saturday we’ll make big changes” and then have the law go into effect… and those changes are not publicly announced anywhere.
This ruling may not mean very much at all. The court has ordered both sides to brief a preliminary injunction over the next couple weeks, with a hearing on August 19th that can get into the actual First Amendment problems with the law. That also means xAI will likely have to explain, in those filings, whatever restrictions it has or hasn’t added to its systems.
xAI’s initial filing was not bad, but I hope they lean more heavily on the case I discussed in my last post, in which Minnesota’s Supreme Court spelled out exactly what the state’s non-consensual intimate imagery law needed to survive strict scrutiny. Because that case walks through, in great detail, the steps a (somewhat similar) law had to take to pass strict scrutiny and survive the First Amendment — even as the court acknowledged the law was punishing a form of protected speech.
There remain some oddities around this law, starting with the fact that the ACLU of Minnesota backed it in the first place — a surprising stance for the ACLU — before reversing course once Musk sued:
xAI’s criticism of the law is drawing support from some free-speech advocates including the American Civil Liberties Union of Minnesota, which supported passage of the law in an earlier form in February but has since turned against it.
“While we believe that creating the technology to alter or ‘nudify’ photos of identifiable people is protected by the First Amendment, we also recognize that the non-consensual creation and dissemination of such material can inflict damage on people appearing in those images,” the ACLU of Minnesota said in a statement Friday.
“In engaging with lawmakers on this issue, we hoped to strike an appropriate balance between First Amendment rights and the ability of people harmed to seek remedial measures, not unlike the remedies available to people harmed by defamation. The final version of the bill does not strike that balance,” the organization said.
Seems like the sort of thing you should have worked out before supporting the law, but fine.
Also, I had mentioned in my last article that some believe the law was written so badly on purpose, to convince Musk to sue in order for Democrats to use it as a political tool and… they are certainly making political hay of it on X, where they seem to be overjoyed that they can mock Elon.
And, sure, mocking Elon is fun. But if you’re going to mock him, it should be over the things he’s actually doing that are problematic.
The issue with this law isn’t that it’s trying to deal with the issue of nudify apps. Or that it’s trying to stop Elon from doing terrible things. It’s that it did so in such a ham-fisted, damaging, obviously unconstitutional manner that bans way more than it claims, is not narrowly targeted, and pretty clearly cannot survive strict scrutiny. Minnesota had the roadmap to pass a legitimate version of this law. It chose not to do so. That Musk didn’t receive the immediate TRO due to the late filing doesn’t make the law a good law. The proper thing for Minnesota’s legislature to do would be to write a law that actually abides by the First Amendment, but perhaps that wouldn’t get them the kinds of headlines they’re getting now.
There’s been a bunch of news this week regarding Minnesota’s new law that purports to prohibit “nudification” technology, and the fact that xAI has sued to have the law blocked as unconstitutional. A few things need to be said upfront, because it’s very, very easy to just say the tech is terrible, that Elon Musk and Grok are terrible, and that of course Minnesota should ban it. But it’s also possible that, in the rush to attack very problematic apps built by very problematic people, Minnesota drafted a bad law that is ridiculously overbroad and pretty clearly unconstitutional. And… that is exactly what appears to be the case.
Let’s start with the basics: apps (mostly powered by various AI tools) that are used to produce modified imagery, especially stripping people of their clothes are… bad. They should be socially shunned. People using them to objectify or sexualize others are doing bad things, and people should judge those who use those apps accordingly. This is not a defense of those apps. Similarly, Elon Musk’s Grok and its widely promoted use of putting people (including children) in bathing suits definitely deserves social shunning as well. Norms take time to form, and the shunning here is still catching up to the technology.
But passing a badly drafted, obviously unconstitutional law does not help form those norms. Nor does it punish Elon Musk. Instead, it allows him to act like a First Amendment martyr.
It’s also worth clearing something up early, because a lot of the coverage has gotten it wrong: this is not a law about child sexual abuse material. CSAM is already quite illegal under both state and federal law, and nothing in HF 1606 is limited to images of minors. Had Minnesota drafted a law narrowly targeting AI-generated CSAM, it might have survived a constitutional challenge. That’s not what it did.
And if you want to pass a law to ban technology like this, there are rules under the First Amendment. And, in Minnesota, we even know what some of those rules are. After all, a decade ago, the state also passed a law criminalizing the dissemination of “nonconsensual private sexual images.” After some back and forth in the courts, the Minnesota Supreme Court finally blessed the law as constitutional in late 2020, but made it quite clear that the law went right up to the First Amendment line. It first noted that while the state wanted to claim there’s an entirely new category of unprotected speech (in this case, “substantial invasions of privacy”), the court refused to do so, citing the famed US v. Stevens case (about an attempt to outlaw animal “crush” videos) in which the Supreme Court made it quite clear that it wasn’t open to creating new categories of unprotected speech:
The United States Supreme Court has emphatically rejected “freewheeling” attempts “to declare new categories of speech outside the scope of the First Amendment.” Stevens, 559 U.S. at 472; see also Jorgenson, 946 N.W.2d at 604 (“The United States Supreme Court has been reluctant to expand these traditional categories of unprotected speech.”). It is possible, however, there are “some categories of speech that have been historically unprotected, but have not yet been specifically identified or discussed.” Stevens, 559 U.S. at 472.
To successfully argue for a new unprotected category of speech, the proponent must present “persuasive evidence that a novel restriction on content is part of a long (if heretofore unrecognized) tradition of proscription.” Brown v. Ent. Merchs. Ass’n, 564 U.S. 786, 792 (2011). This is a heavy burden to bear, and the Supreme Court has recently rejected creating new categories of unprotected speech for animal cruelty, Stevens, 559 U.S. at 472, depictions of excessive violence, Brown, 564 U.S. at 791–93, and false statements, Alvarez, 567 U.S. at 722–23.
In this case, we conclude that the State has failed to carry the heavy burden required to provide a basis to establish a new category of unprotected speech.
And yet, the law was still deemed constitutional, but not because it created a new category of unprotected speech, but rather because it passed strict scrutiny, in which the law is narrowly tailored to use “the least restrictive means” of addressing a compelling government interest. That is the test by which a law can still be deemed viable under the First Amendment, despite suppressing speech. In the case of the nonconsensual imagery bill, the law passed strict scrutiny because it focused very narrowly on a category of speech that is very likely to cause harm, and put in place a law that was narrowly tailored to only target that speech, and on top of that included clear exemptions for edge cases that likely wouldn’t be harmful.
Indeed, the court leaned hard on the fact that the law only reached images disseminated without consent, and only when the disseminator knew or reasonably should have known the subject expected privacy. Those two limits — consent and intent — are what kept the statute from sweeping in vast amounts of protected speech. Some quotes from the court which list out all the factors necessary to pass strict scrutiny.
First, the Legislature explicitly defined the type of image that is criminalized…. Furthermore, the image has to be “obtained or created under circumstances in which the actor knew or reasonably should have known the person depicted had a reasonable expectation of privacy.” Id., subd. 1(3). Images that do not clear each of these hurdles fall outside the scope of the statute.
Second, a defendant must “intentionally” disseminate the image. … This mens rea requirement means that a defendant must knowingly and voluntarily disseminate a private sexual image; negligent, accidental, or even reckless distributions are not proscribed. This specific intent requirement further narrows the statute and keeps it from “target[ing] broad categories of speech.”
Third, the statute has seven enumerated exemptions…. The statute allows for private sexual images to be distributed “in the course of seeking or receiving medical or mental health treatment.” Id., subd. 5(3). Advertisers, booksellers, and artists are protected because images “obtained in a commercial setting” for legal purposes fall outside the statute’s reach. Id., subd. 5(4). Journalists cannot be prosecuted because there are exemptions for the dissemination of private sexual images that involve matters of public interest and “exposure[s] in public.” Id., subd. 5(4)–(5).8 Educators and scientists are protected because there is an exemption for private sexual images disseminated for “legitimate scientific research or educational purposes.” Id., subd. 5(6). Accordingly, even if protected speech falls within the ambit of subdivision one and a disseminator acted with the requisite mens rea, that person may still be exempt from prosecution under these precise exceptions.
Fourth, to be prosecuted under the statute, a disseminator must act without consent…. This provision provides additional protection for commercial advertisements, certain adult films, artistic works, and other creative expression outside the statute’s scope.
Finally, this statute only encompasses private speech…. Unlike the overly broad statutes at issue in our recent decisions in In re Welfare of A.J.B. and Jorgenson, this statute covers only private sexual images and does not prohibit speech that is “at the core of protected First Amendment speech.”
It was all of that combined that allowed the law to pass strict scrutiny — something that is incredibly difficult to do. Most laws that have to clear strict scrutiny don’t. Here, this law survived with a careful roadmap from the court of how to do so.
One would think that Minnesota legislators would be aware of this ruling and the clear reasons why the law was deemed to pass strict scrutiny and then write an equivalent law with the same elements in trying to ban nudify apps.
But for reasons known only to the Minnesota legislators, they basically ignored every single one of those points.
Minnesota’s anti-nudification tech law is not limited to non-consensual content. This means, as legal commentator Kathryn Tewson noted, that if she uploaded a picture of herself and asked Grok to put her in a bikini, she could by her own hand, cause Grok to break this law. That… seems like a very problematic law.
See, as I read this law, if I uploaded a picture of myself in a sundress and said “Grok, make this a picture of me in a bikini instead,” it would be a violation of the law for Grok to do that. I don’t think that should be illegal.
And, again, the Minnesota Supreme Court has already told the state pretty much exactly how to make this law constitutional: focus on nonconsensual imagery, narrowly tailor it to just the deeply harmful content, include an intent requirement, and include clear delineated exemptions for things that should be allowed.
Minnesota legislators did none of that. Indeed, even the definition of “intimate parts” in the law borrows its definition of ‘intimate parts’ from an earlier statute, covering: “the primary genital area, groin, inner thigh, buttocks, or breast of a human being” — not much of which is inherently sexual, let alone harmful. Tewson offers another example: an edit of a Taylor Swift photo that changes the texture of her fishnet stockings to look more like skin. Whatever tool made that edit just violated Minnesota law.
2. An image generated from this image of Taylor Swift performing on the Eras tour in which the texture of her legs in the modified image appears more similar to actual skin than it does to nude-fishnets-over-nude-tights:
This is, by definition, an overly broad, non-narrowly tailored law.
Another example: last year the TV show South Park did a deepfake parody of Donald Trump, showing a photorealistic version of him wandering naked through the desert, including his “intimate parts.”
Under this law, that video could violate HF 1606. That’s not narrowly tailored. That’s not dealing with intent or focused just on truly harmful content.
One lawyer I spoke to, after reading through the statute, wondered out loud whether the Minnesota legislature had deliberately drafted it in the dumbest way possible just to guarantee a successful challenge. That’s how poorly the law was drafted.
Of course, no one wants to hear that the law is badly drafted. Lots of people want to ban nudify apps and to yell about how ridiculous it is that Elon Musk has gone to court to challenge this law.
But… it’s the sort of thing he should be doing. Otherwise anyone can have Grok put themselves in a bikini and… Minnesota’s Attorney General can demand $500,000 for each such image created, even when the image was created deliberately, by the person in it, of themselves.
xAI (now a division of SpaceX) is right to challenge the law, not because nudify apps are a good thing, but because the law is terribly drafted and pretty clearly exceeds what’s allowed under the First Amendment. The complaint itself is worth a read. For one thing, it explains why xAI last week sued one of its own users for producing CSAM with Grok (which I had found perplexing at the time). It reads a lot like the company wanted a concrete example to put in this filing of how it fights back against those who use Grok in such ways (leaving out, of course, that Elon himself used the app to put himself in a bikini, thereby encouraging others to do the same).
It also explains why that complaint was focused on triggering the indemnity clause in X’s terms of service, which makes the user liable for any legal costs associated with their use of the product. What Musk is really signalling with that lawsuit is if Minnesota’s AG sues us under this law for your usage of the product, we’re going to sue you to cover our costs (which could include the $500,000 fine for any images created).
As the lawsuit notes, the law is just terribly written:
HF 1606 punishes AI platforms that allow users to alter images of real people to depict an “intimate part.” But the statute contains no knowledge, intent, or purpose requirement. It is a strict-liability statute keyed solely to whether a user succeeded in creating a covered image using the AI provider’s platform—regardless of whether the provider prohibits users from using its tool for such a purpose, regardless of how many mitigations the provider has in place, and regardless of how diligently the provider polices such conduct using its tool. There is no safe harbor for good-faith efforts of the provider of general-purpose AI creative tools to avoid harms. Liability attaches even if the depicted persons consented—or created the image themselves—and even if the image is never shared. Liability also attaches even if the image has artistic, scientific, political, satirical, educational, medical, or religious value, and (again) even if the company has deployed near-perfect, state-of-the-art technical controls to prevent the generation of nude images.
Additionally, the law’s definition of “intimate part” is exceptionally broad. Although the federal government and various states have enacted statutes that clearly define nudity for the specific context of AI-generated images, Minnesota rejected such a precise definition. Instead, it borrowed the definition of “intimate part” from a criminal sexual-contact statute. That definition was drafted for nonconsensual touching and thus covers the inner thigh, buttocks, or breast of a man or woman, as well as the groin and primary genital area. HF 1606 accordingly bans ordinary depictions of men without shirts, people in shorts or swimsuits, and other body parts routinely displayed in public—far beyond what an ordinary person would consider “nudification.”
Even worse, as the lawsuit states, the bill’s “principal sponsor” admitted that the law was designed to apply to consensual imagery:
A service used by an adult to edit a photograph of him or herself or a consenting individual is covered on the same terms as a service used to create an image of an unwilling stranger. The statute’s text draws no distinction among them. And this was by design. When a staff member of the Senate Judiciary and Public Safety Committee pointed out that the Act’s “prohibition applies to consensual images,” Senator Maye Quade (the bill’s principal sponsor) explained “that is intentional.”
That is the bill’s main sponsor stating, on the record, that she deliberately chose to leave out one of the very features Minnesota’s own Supreme Court had identified as necessary for a law like this to survive constitutional scrutiny.
That is legislative malpractice.
Since the lawsuit was filed, Maye Quade and other legislators have publicly defended the bill:
“I don’t see this as a free speech issue. This does not regulate content; it does not regulate art. It regulates conduct,” Maye Quade said. “Prompts are not art, and we protect art specifically in this law. It’s pretty audacious to sue to prevent a law that protects children from being turned into child sexual abuse material.”
She’s describing a law she could have written, but didn’t.
Notice what’s missing from that defense: any explanation of why the consent and intent elements the Minnesota Supreme Court specifically identified as saving the 2016 law were left out of this one. Also, she’s just simply incorrect that the law does not regulate speech. Again, if she simply read what the Minnesota Supreme Court said about the nonconsensual intimate imagery law, it spent pages analyzing the nonconsensual imagery statute — a law covering narrower material than this one — as a content-based restriction on speech that had to pass strict scrutiny to survive.
Similarly, the law does not actually “protect art.” Its one and only exemption is if the work “requires the technical skill of a user to nudify an image or video.” That could protect some art, but not all. And it defines art only in a case where a level of skill is needed, which itself potentially creates First Amendment issues in defining what is, and what is not art. There is plenty of modern art that people regularly complain takes no “technical skill” to create.
The complaint itself includes some other examples of what would violate the law, including this (gross) AI-generated image that Trump posted of a slimmed down version of himself, some of his cabinet members, and a randomly generated woman in a bikini sitting in a gleaming blue reflecting pool. Under the law, whatever tool was used to generate that image pretty clearly violated Minnesota’s law:
In this viral snapshot—which President Trump posted publicly— President Trump, Vice President J.D. Vance, Secretary of State Marco Rubio, and Secretary of the Interior Doug Burgum all are portrayed shirtless in the Washington Mall’s reflecting pool, along with an unknown (possibly fictitious) woman.19 An “intimate part” (the breast) of at least the President, Vice President, Secretary of the Interior, and the woman are “depict[ed],” with the Secretary of State also at least arguably included as well. The President posted this image on his personal account, presumably to make light of the public controversy surrounding repairs to the reflecting pool on the National Mall.
Nudify apps are gross. Musk’s encouragement of people to use Grok to de-clothe people is gross. People who use AI tools to “nudify” people are gross. But that doesn’t mean all laws targeting such things are good laws or constitutional.
In this case, despite having clear instructions from its own Supreme Court on how to write a constitutional law, Minnesota’s legislature deliberately chose to write an unconstitutional one. And thus, this lawsuit is the proper thing for SpaceX/xAI/Musk to do.
Supporting the lawsuit is not supporting Elon or Grok or nudify apps. It’s telling every legislature in the country the same thing: if you want the law to survive, learn to draft it in ways that aren’t unconstitutional.
You may recall that at the end of last year and very beginning of this year there was a big story over Elon Musk’s LLM tool Grok being willing to produce non-consensual images of people (including children) in bikinis or otherwise stripped down in objectifying ways. Elon played into and promoted this “trend” at one point by asking Grok to modify an image someone had already asked Grok for putting a bikini on the famous Ben Affleck smoking meme, asking it to replace Affleck’s head with Musk’s own. Grok complied and while the resulting image has since been removed, Musk responded “perfect.”
While some will argue this is more acceptable, as it was Musk doing it to himself, it still generated a ton of extra attention to the idea that Grok would gladly, publicly, undress just about anyone and put them in a bikini. To Musk, apparently, it all seemed like good fun.
As more and more of the media pointed out that this was being done to children, in some cases potentially creating illegal child sexual abuse material, Musk eventually came out with a statement claiming that anyone using Grok that way will “suffer consequences.”
In the interim, a number of people have been arrested for using Grok to produce CSAM, in some cases based on X’s reports to NCMEC. Of course, the company is also being sued by a bunch of people, claiming that it should be liable for the non-consensual imagery that it created. Some of the details of the lawsuit linked above are particularly horrifying:
Jane Doe 4, a female from Wyoming, said her stepfather uploaded a photo of her when she was 11 and lying on a couch to his phone. Using Grok, the stepfather created more than 7,000 CSAM-related images of her. He also shared and traded the images with others on social media platforms.
The lawsuit alleges that the stepfather opted for Grok “because the platform was less restrictive than other AI models and responded to his prompts to generate sexually explicit material using an image depicting a prepubescent minor.”
It also claims that in February, xAI did generate a tip to the National Center for Missing and Exploited Children regarding the images, but the company only submitted the original, authentic image as evidence. According to the suit, xAI did not respond when law enforcement requested the thousands of Grok-generated images based on the photo and IP address information that would have quickly helped identify her stepfather as the perpetrator.
Perhaps recognizing that this story is getting worse, xAI (the parent company of X since it acquired the social network in yet another deal of Elon “selling” one of his companies to another of his companies) has now directly sued one of its own users, Terry Harwood, for creating CSAM with Grok, arguing that the indemnification clause in the terms of service means that the court should order Harwood to pay for xAI facing lawsuits from Harwood’s victims.
The lawsuit, filed in Musk’s favorite federal court — the Wichita Falls Division of the Northern District of Texas — was immediately assigned to Musk’s favorite judge, Reed O’Connor. It sure seems like the company is trying to get out ahead of this story before it gets even worse — by suing one of its own users for doing, in a horrific and illegal way, exactly the kind of thing Musk himself had promoted.
The filing, of course, tries to play up the claim that X is a good actor in this space, taking its responsibility seriously:
There are bad people in the world and not all users have good intentions, so Plaintiff draws a bright line on permissible usage: the tool may be open and truth-seeking but is not a safe harbor for illegal, harmful, or abusive conduct. Plaintiff takes violations of this rule extremely seriously. It enforces its rules against violators through account suspensions, account terminations, and by reporting suspected child sexual abuse material to the National Center for Missing & Exploited Children (“NCMEC”). Indeed, Plaintiff has suspended 52,222 accounts and made 73,604 reports to NCMEC in 2026, resulting in (at least) 244 arrests.
In addition and when necessary, Plaintiff will pursue civil accountability against bad actors who abuse its tool to harm others.
Plaintiff strictly prohibits any use of Grok in connection with illegal, harmful, or abusive activities. These prohibitions are set forth in plain language in the xAI Terms of Service and the xAI Acceptable Use Policy. Among the expressly forbidden activities are:
“Using [Grok] in a way that infringes, misappropriates or violates a person’s privacy or their right to publicity”;
“Undressing or nudifying real persons, or otherwise altering a real person’s image or likeness to depict them in an intimate or sexual context”;
“Depicting likenesses of persons in a pornographic manner”; and
“Sexualizing or exploiting children.”
Which all sounds well and good if you simply ignore the whole “yeah, but your CEO kinda laughed about and promoted this use of the tool” part.
Also, it’s quite a move to sue a user for using a tool that you offer in a manner that you enabled the tool to work. I’ve long argued that users themselves are the ones who should be liable for any violative actions they make rather than the platform but I don’t recall ever seeing a platform actually sue one of its own users for using their own tools. It really feels like them looking at all the lawsuits coming their direction for Grok’s usage and… pulling the whole “we’re all looking for the guy who did this” hotdog costume meme in the form of a lawsuit.
I have no idea if it’ll work, but suing your own users for the very thing that other people are suing you over is quite a statement. Yes, in this case it involves potentially criminal behavior (the guy was arrested after all), but are Musk and X suggesting that if you violate any part of their terms of service, Musk can take you to court in front of his favorite judge? Because it sure seems to suggest that. Much of the argument is based on laying out X’s Terms of Service and Acceptable Use Policy and detailing why this guy violated them. This kind of argument does not at all appear to be limited to law-breaking activity:
Defendant breached the xAI Terms of Service and Acceptable Use Policy by leveraging Grok to generate non-consensual sexually explicit images and CSAM.
Among Defendant’s various other breaches described herein:
a) Defendant manipulated Grok to create sexually explicit images of others without their consent, in contravention of the xAI Terms of Service, which prohibit, among other conduct, using Grok to aid in violating a person’s privacy or their right to publicity.
b) Defendant manipulated Grok to create CSAM, in contravention of the xAI Terms of Service, which prohibit, among other conduct, the sexualization or exploitation of children.
Plaintiff has been damaged by Defendant’s breaches of the xAI Terms of Service.
The lawsuit leans heavily on the indemnification clause in their terms of service to argue that they want the court to order Harwood to cover any legal expenses that might stem from potential lawsuits from the people Harwood victimized with his CSAM.
The xAI Terms of Service contain a clear and conspicuous indemnification provision that states: “You agree to indemnify, defend, and hold harmless xAI, its affiliates, officers, directors, employees, and agents from and against any and all claims, demands, losses, damages, liabilities, costs, and expenses (including reasonable attorneys’ fees and court costs) arising out of or in any way related to (a) your access to or use of the Services, (b) your User Content or Outputs, (c) your violation of these Terms or any applicable law or regulation, or (d) any third-party claim resulting therefrom, including but not limited to claims of privacy violation, right-of-publicity infringement, or harm caused by content you generate or distribute using the Services.”
Defendant’s conduct directly triggered this indemnity obligation. As detailed above, Defendant knowingly and intentionally used Grok to circumvent safeguards, alter nonconsensual images, and generate and distribute CSAM and NCII.
There is no indication as far as I can tell that Harwood’s victim(s) have taken legal action against xAI at this point, so it’s a little unclear what they’re hoping to get out of the indemnification here other than to present themselves as taking a (somewhat late) stance against this kind of illegal usage of a tool that they failed to properly lock down while wink-winking at how they expected people to use the tool.
Harwood, assuming the allegations and criminal charges hold up, did something genuinely monstrous, and his victims deserve real justice. But it does feel very wrong that the same company whose CEO demonstrated this use of Grok on himself for laughs, and whose product made this kind of abuse trivially easy to produce at scale, is now standing in front of Musk’s favorite judge asking to make one user personally responsible for some of the fallout — while the “we built and promoted the machine” part stays out of the complaint entirely.
And, honestly, given how much Musk has promoted X as “the anything app” where you can do and say anything, it does seem notable that he’s now suing a user for… doing exactly that.
Last week Elon Musk successfully conned America and U.S. regulators into signing off on his preposterous SpaceX IPO, which immediately generated Musk $75 billion by comically over-stating the value of SpaceX, xAI, and Starlink. Then bone-grafting the entire pile of bullshit to the U.S. economy and your retirement account under the pretense that space data centers and Mars colonization are just around the corner.
A handful of remaining useful journalists have repeatedly explained how xAI and Musk’s racist 5th place chatbot — which comprises the lion’s share of the ridiculous IPO valuation — is a gargantuan loser. Both SpaceX and xAI aren’t profitable and may never be, and the claims of Mars colonization and space data centers are unworkable bullshit designed to distract people with toddler-level critical thinking skills.
Anyway I’m sure it will go fine.
As a multi-decade telecom beat reporter I’d say I’m better positioned to talk about Starlink — the only actually profitable company in the SpaceX IPO prospectus (and that’s assuming Starlink is being honest about their financial numbers in a country too corrupt to have working financial regulators).
I’ve long noted how Starlink is great for people with no other options, but data has shown how it’s too congested to meaningfully scale. It’s also often too expensive for the sorts of Americans struggling with access. There’s also the problem with it ruining astronomical research and degrading the ozone layer. So Starlink is great for RVs or a guy with an extra cabin in the woods, but it’s not a miracle.
In terms of broadband policy, it’s supposed to be a niche solution. The kind of technology you use to fill in the gaps after you’ve pushed fiber, 5G, and fixed wireless out as far as you can into unserved areas.
But as I’ve mentioned previously, folks in the Trump administration and extended Rogan infotainment universe see Starlink as akin to magic. They think it’s just a sort of pixie dust you sprinkle over the entire of U.S. connectivity woes. There was a soggy Bulwark interview last week with Jason Calacanis that kind of reveals how deep the delusion goes in terms of what Starlink actually is:
The SpaceX IPO insists — and Calacanis dutifully believes — that it’s trivial for Starlink to jump from a niche satellite broadband solution with a little over 10 million subscribers to a massive economic powerhouse with 300-500 million subscribers. Calacanis waxes poetic about Starlink providing bandwidth to every phone in the world and surpassing even Netflix in terms of total subscribers.
But in a way that’s highly representative of modern Silicon Valley, Calacanis doesn’t actually care about how the tech works, or even if it works. Calacanis is interested in unchecked wealth accumulation, and propping up the unbridled profit-seeking of a personal friend.
The thing is: to meaningfully grow, Starlink will need to start seriously competing on price to counter competitors (like Amazon) coming into the space. But the cost of endlessly replacing LEO (low Earth orbit satellites) is immense (SpaceX says each satellite has a five year lifespan, but it’s arguably much lower). And ARPU is already dropping for Starlink as the company tries to drum up new subscribers.
Calacanis insists Starlink’s just a hop, skip, and a jump from being even bigger than Netflix. But for Starlink to even sniff those kinds of numbers, it would have to intensely compete with deeply-entrenched and politically-powerful telecom monopolies, and fiber optic broadband and 5G/6G networks less constrained by the rules of physics. They’ve also got to compete with a rising tide of community-owned fiber.
As Starlink grows its subscriber base, it’s not only going to see its ARPU drop faster, but data shows it’s going to run into new capacity constraints. That means more annoying network management practices that throttle video, limit services, and generally degrade performance. We’re already starting to see the impact of this with network slowdowns and “congestion fees” ranging upwards of $750 in some areas.
Anybody claiming that Starlink is the ticket to vast riches is either lying to you or doesn’t understand how the technology actually works. Even if it can maintain its success as a viable niche connectivity option useful in rural markets and global battlezones, the high cost of maintenance means this is never going to be a major money maker. Though they clearly hope it will prove to be a semi-useful backbone for a major pump and dump scheme.
The ace Elon Musk is holding is corruption and cronyism leading to regulatory favors and massive new subsidies, but it’s not clear even that’s going to be enough.
Cecilia Kang at the New York Times has an interesting article about how the Trump FCC has been doing cartwheels trying to prop up the Musk IPO — especially as it pertains to Starlink. That has included not just abandoning any meaningful regulatory oversight of “space junk” and orbital safety, but launching dodgy investigations into companies that hold spectrum Musk wants for himself.
Elon Musk bought himself a Presidency, and it continues to pay off handsomely:
“Carr has taken multiple actions for which Musk was the prime beneficiary,” said Blair Levin, an adviser to New Street Research, an investment research firm, and a former chief of staff at the F.C.C. He added that Starlink “has gotten a huge amount from the Trump administration and Carr.”
Carr has tried to justify his favoritism of Musk by saying he’s also rubber stamped the LEO satellite policy interests of Jeff Bezos and Amazon. But as we’ve consistently established around here, nothing Carr does is driven by any sort of good faith concern about the public interest.
The funny part is that the New York Times doesn’t even mention that the Trump administration has also hijacked the 2021 infrastructure bill to redirect potentially billions of dollars to Elon Musk and Jeff Bezos (I should have an upcoming feature on this over at The Verge). This is money being directed away from affordable fiber and toward two billionaires — for networks they already planned to build.
More specifically, the Trump NTIA under former Ted Cruz staffer Arielle Roth changed the language of the $42.5 billion Broadband, Equity, Access, and Deployment (BEAD) program so that Musk and Bezos would be the prime beneficiaries. They also stripped out any language requiring that internet access built with taxpayer money had to be affordable or equitably deployed with an eye on fairness.
Musk and Calacanis types try to brush functional oversight for taxpayer spending as unnecessary “wokeness.” But the ongoing BEAD saga involves an historic hijacking of Congressionally-mandated funds by bad faith actors; so it’s curious the New York Times didn’t think it was worth mentioning in a story about how unethically cozy the Trump administration and Musk are.
Like most of the SpaceX IPO this will all be proven out over time. Long after people have had their retirements account raided, or small towns have had their infrastructure hopes hijacked. Consumers, taxpayers, and labor will, as is usually the case, be left holding the bag. And the folks that made it possible will already be off to the next big thing leaving people of conscience to clean up the mess.
Elon Musk, business genius. When Elon Musk announced his plans to buy Twitter, some of his billionaire friends rushed to text him to say they’d throw whatever money they wanted into the deal. Larry Ellison casually offered “a billion… or whatever you recommend.” Marc Andreessen offered $250 million, no questions asked. This all came out in the lawsuit when Musk tried to back out of the deal:
Publicly, these billionaires insisted that Elon was a sure shot business genius who would easily make them much richer. Elon then sent around a presentation to other investors who would perhaps take a bit more convincing. The NY Times got its hands on Elon’s clearly pulled-out-of-his-ass projections. $26.4 billion in revenue by 2028! That included $12 billion from advertising, $10 billion from subscriptions and the rest from licensing.
Remember, at the time, Twitter’s ad revenue was decent: $4.51 billion in 2021 (its last full year as a public company) with another half a billion in licensing revenue. So Elon was suggesting he had the magic formula for massively increasing ad revenue and subscription revenue.
There was plenty of reporting over the last few years on how the opposite happened. Ad revenue absolutely tanked. It got so bad that the company started suing advertisers for not advertising on the newly renamed X (and threatening advertisers that choosing not to advertise would get them added to the lawsuit), pretending that it was some sort of antitrust violation. It took a court to point out that this was utter nonsense.
Anyway, given the private nature of X, we didn’t have any real official confirmation on some of the revenue numbers. But in the last year and a half, Elon has been merging his Xs. He merged X into xAI, then merged xAI into SpaceX. And now SpaceX has filed for a massive IPO, giving us an S1 with some financial information about how X is actually performing after all.
Of course, by merging all these companies, it gives Elon a bit of a chance to obfuscate the numbers. The user metrics, for example, show both users of X and xAI’s grok (which are not all the same). Also, somewhat ironically given Elon’s pretextual whining about how there were too many bots on Twitter, the S1 admits that a lot of the activity on X these days is almost certainly bots and they apparently have no way to break out how many humans still use the service:
“supported accounts” refers to, when used in the context of our X platform and Grok, a human, bot or similar account that logged into the X platform or Grok. The total number of supported accounts may include fake, spam or bot accounts if they are active.
Gosh. I thought you were taking over the site to get rid of all the bots and spam.
Anyhoo, now that we have some numbers, let’s compare them to what Elon sold his investors.
Remember, the plan was $26.4 billion by 2028. We’re more than halfway there. How’s it going? Well… when he combines xAI (grok) revenue with X revenue (so not even just breaking out X’s ad revenue)… we get… a total of $3.201 billion in 2025. So, just to put this in perspective… when he took over in 2022 he laid out a five year plan to take the company that had $4.5 billion in ad revenue the year before he bought it up to $12 billion in five years. Three years in and… it’s now somewhere pretty far below $3 billion. And they’re proud of the fact it’s finally started to go up again:
Revenue for the year ended December 31, 2025 increased by $581 million, or 22.2%, compared to the prior year ended December 31, 2024. This increase was primarily due to an increase in advertising revenue of $116 million as advertising spend increased from advertising partners on X and an increase in AI solutions and infrastructure revenue of $465 million.
So… from 2024 to 2025… they increased advertising revenue on X… by… $116 million, after knocking it down by somewhere in the range of $2 billion? BUSINESS GENIUS.
But, that’s okay. Part of the pitch was that he was going to get advertising to be less than 50% of Twitter’s revenue by 2028 because it was going to be replaced by a massive wave of subscription revenue. $10 billion by 2028! Musk predicted 69 million users of Twitter Blue (what became X Premium) by 2025 and 159 million in 2028. And then also another 104 million subscribers to a mysterious “X” subscription by 2028, which was not explained in the pitch. Even though this was before the rollout of ChatGPT, if we want to grant Elon credit to think he had already planned to launch an AI subscription service called “X” by then… how are we doing towards those numbers?
As of March 31, 2026, we reached approximately 6.3 million active paid subscribers, which was comprised of approximately 4.4 million X Premium and Premium+ paid subscribers and approximately 1.9 million SuperGrok, SuperGrok Heavy and SuperGrok Lite paid subscribers.
Leaving aside the Grok subscribers… they have… 4.4 million X Premium subscribers. That seems a bit short of the 69 million paid subscribers (which was almost certainly chosen because Musk is, emotionally, a 12-year-old boy). Once you combine that with the Grok subscribers (most of those plans cost significantly more than X Premium) and you get a grand total of… $365 million. Given the breakdown of X vs. Grok subscribers and the different pricing, X subscribers likely account for less than two-thirds of that revenue — call it under $250 million. That seems juuuuust a bit short of $10 billion.
His initial pitch to investors also projected that by 2028 the payments business would be bringing in over a billion dollars. It’s now 2025 and while the S1 mentions payments, it’s very much a future thing:
We plan to further broaden the value proposition of X through offerings like Money, a product we launched in beta in November 2025, which aims to expand platform utility by enabling payments and other financial services.
In the pitch to investors, the plan was to have that generating revenue by 2023. A bit behind schedule, it seems.
Also, part of the pitch was that all the debt he’d taken on would be paid back through free cash flow. He even says that by 2025 (hmm… last year…) the company would grow to $3.2 billion. Uh, not so close. Again, that almost matches the revenue number, but the cash flow was… decidedly negative. The entire AI part of the business lost over $6 billion last year. I don’t think Elon’s paying off the debt with free cash flow any time soon.
Look, obviously, forward looking projections and investor pitches are fantasies. They always are. That’s kind of the point. And also, obviously, the consumer AI/LLM race which really became a consumer phenomenon started right after Musk closed the purchase, and shifted the landscape somewhat. Also, obviously, by merging X into xAI and then merging that combined company into SpaceX, the various investors are likely to make out just fine (even if it is stacking multiple houses of cards on top of each other).
But, given how there was a group of Silicon Valley VCs and Wall Street banker types who absolutely insisted that Elon had a Midas touch and would absolutely know how to turn Twitter into revenue gold, it seems worth checking in on just how badly those plans failed. Yes, he’s been able to paper that over with mergers between companies he owns, but the actual numbers don’t lie.
So where does this leave the investors who lined up to hand Elon a few billion dollars, no questions asked? Probably fine, actually. The SpaceX IPO will almost certainly value the combined entity at a number that makes early Twitter/X investors more than whole. That’s what merging a struggling social network into a so-so AI startup into a deeply in debt (but in strong demand) rocket company will get you — the underlying failure gets laundered by the valuation of everything else in the stack.
But the operational track record is what it is. Twitter was generating $4.5 billion in ad revenue the year before Musk bought it. Three years into his five-year plan to reach $12 billion, the combined X/xAI advertising business is at somewhere under $3 billion — and that’s counting the separate AI business he launched after acquisition. The 69 million paid subscribers became 4.4 million. The $10 billion subscription business became $250 million. The payments business that was supposed to be generating revenue in 2023 just launched in beta in November 2025.
The “business genius” narrative was always doing a lot of work. Now we have the numbers. They don’t.
Remember when the Biden administration set up something called the “Disinformation Governance Board” and the entire MAGA universe lost its collective mind? It was the “Ministry of Truth.” It was “government speech police.” It was the single most Orwellian thing any American administration had ever done in the history of civilization. Nina Jankowicz, the researcher tapped to lead it, received death threats. The whole thing was shut down within weeks because of the outcry.
Of course, all of it was an exaggeration. That board was actually set up to coordinate efforts to counter foreign disinformation — not to police Americans’ speech. We said so at the time, even while criticizing DHS for the monumentally stupid way they named and rolled it out. The name was terrible. The communication around it was worse. But the underlying mission — helping coordinate the government’s own efforts to respond to (not censor) foreign influence operations — was legitimate and, frankly, important in this era of information warfare.
Well, Secretary of State Marco Rubio just signed a cable doing something that sounds vaguely similar, but way worse. Specifically, he’s directing U.S. embassies and consulates worldwide to launch coordinated campaigns countering foreign propaganda — and the cable explicitly endorses Elon Musk’s X as an “innovative” tool for the effort. It also admits that this is pure psyops work:
The cable instructs those embassies and consulates to pursue five broad goals: countering hostile messaging, expanding access to information, exposing adversary behavior, elevating local voices who support American interests, and promoting what it calls “telling America’s story”. Embassies are told to recruit local influencers, academics and community leaders abroad to carry counter-propaganda messaging, an approach designed to make American-funded narratives feel locally organic rather than centrally directed.
“These campaigns seek to shift blame to the United States, sow division among allies, promote alternative worldviews antithetical to America’s interests, and even undermine American economic interests and political freedoms,” the cable says. “Using digital platforms, state-controlled media, and influence operations, they pose a direct threat to US national security and fuel hostility toward American interests.”
Notably, the cable tells diplomatic offices to coordinate their work with “the Department of War’s Psychological Operations” – the military unit more commonly known as Miso, or Military Information Support Operations, formerly Psyop, which is part of the Pentagon.
This is far more expansive than anything the Disinformation Governance Board ever even contemplated — and the same people who screamed about the Ministry of Truth are, once again, completely silent.
The idea that the State Department would issue a formal cable endorsing a specific social media platform by name as a tool of U.S. diplomacy—let alone military psychological operations—would have been, until recently, almost unthinkable. But the structural transformation that has taken place over years has made the news feel almost ordinary today. It was a transformation that dismantled, piece by piece, the legal accountability, operational independence and institutional resilience that once made such a cozy relationship between government and platforms inconceivable.
And see if any of this sounds familiar:
Rubio identifies five operational goals—countering hostile messaging, expanding information access, exposing adversarial behavior, elevating local voices sympathetic to U.S. interests, and “telling America’s story”—and instructs embassies to recruit local influencers and community leaders to carry U.S.-funded narratives in ways designed to feel organically local rather than centrally directed.
Why, that sounds quite similar to what the Biden DHS said about the Disinformation Governance Board. Except, suddenly: no partisan freakout. No weeks of stories on Fox News. No screaming in the NY Post about speech police. Gee. I wonder why.
The U.S. State Department is instructing embassies to recruit local influencers to carry U.S.-funded narratives in ways designed to feel organically local rather than centrally directed. This is, by definition, a covert influence operation. It’s the kind of thing that, when other countries do it, we call propaganda. It’s the kind of thing the Global Engagement Center was specifically designed to expose.
Oh, right. About the Global Engagement Center.
You may recall that one of the early moves of the returning Trump administration was to shut down the GEC, the State Department office specifically created to help identify and counter foreign influence campaigns. At the time, Rubio — the same Marco Rubio who just signed this cable — framed the shutdown as a free speech victory:
Under the previous administration, this office, which cost taxpayers more than $50 million per year, spent millions of dollars to actively silence and censor the voices of Americans they were supposed to be serving. This is antithetical to the very principles we should be upholding and inconceivable it was taking place in America.
That was always a lie. The GEC (just like the Disinformation Governance Board) didn’t “silence and censor” Americans. It studied foreign influence campaigns — the kind run by Russia’s Internet Research Agency, by ISIS recruitment networks, by Chinese state-linked information operations — and helped expose them. It’s the kind of work that requires sustained expertise, institutional knowledge, and sophisticated analytical capacity. The kind of thing you can’t just spin up overnight when you suddenly realize you need it.
So all of the hand-wringing about the Disinformation Governance Board, the GEC, and the idea that governments were too close to social media platforms was a bunch of nonsense all along. It was always about trying to gain and then keep power, destroying the institutions that dealt with foreign disinformation campaigns until they could capture them for their own purposes.
Klonick traces how Twitter/X became susceptible to exactly this kind of capture:
Musk systematically dismantled Twitter’s trust, safety, and content moderation infrastructure. The teams that had worked, however imperfectly, to maintain platform integrity not just for commercial reasons but to limit the spread of coordinated inauthentic behavior, state-linked influence operations, and targeted harassment were gone within months of Musk’s ownership. With both the corporate accountability architecture and the internal operational safeguards stripped away, the platform’s amplification and suppression mechanics became, in effect, tools that could be deployed at anyone’s, but namely Musk’s, discretion.
Before Musk’s acquisition, the major US tech platforms — whatever their flaws — generally bent over backwards to avoid being captured as instruments of state messaging.
The Rubio cable, on the other hand, specifically endorses X’s Community Notes feature as a tool for countering “anti-American propaganda operations without compromising free speech.” Klonick correctly identifies this as:
…a remarkable exercise in circular reasoning: the government endorsing, for use in state-directed information operations, a moderation tool on a platform owned by a former (and perhaps still current) senior government advisor.
But it’s worse than circular reasoning. Community Notes is a crowdsourced system. Its outputs are determined by which users participate and how they coordinate. While it’s (actually very cleverly) designed to avoid brigading attacks, that does not mean it’s perfect in avoiding manipulation. If the U.S. government can organize sympathetic actors to use Community Notes to surface pro-American narratives as part of a formal PSYOP-adjacent campaign, then so can every other government on the planet. China can coordinate its own actors. Russia already runs exactly these kinds of operations. Iran has entire units dedicated to this. The cable essentially advertises to every adversary exactly how to game the system — and the people who actually understood these vulnerabilities, the trust & safety teams, the GEC researchers, the disinformation scholars, are exactly the people this administration spent years attacking and driving out of their jobs.
Oh, unless they expect Elon Musk to tilt the playing field to their advantage — which is exactly the kind of thing these very same people were loudly freaking out about when Biden was president.
Now, some might point out that the broader “censorship industrial complex” crusade wasn’t only about counter-messaging efforts like the DGB and the GEC. It was also about the Murthy v. Missouri case, which dealt with something categorically different: the allegation that the Biden administration pressured platforms to remove third-party users’ speech. The Rubio cable, by contrast, directs government employees themselves to use the platform for their own messaging. These are genuinely different things.
But the supposed animating principle behind the entire crusade was that the government had no business being entangled with social media platforms on matters of information and speech. Not just “the government shouldn’t pressure platforms to remove user content,” but the much broader claim that any government-platform coordination on information amounted to a sinister censorship machine.
Jim Jordan’s “censorship industrial complex” hearings didn’t just target White House communications with platform trust & safety teams. They went after researchers. They went after the GEC. They went after nonprofits studying foreign manipulation. The message was that any institutional involvement in the information ecosystem was inherently suspect. That principle, it turns out, had an expiration date — specifically, January 20, 2025.
And remember, in the Murthy case itself, the Supreme Court rejected the argument that the Biden admin’s communications with platforms constituted coercion. The plaintiffs couldn’t even establish standing because they couldn’t show the government actually changed platform behavior. Meta felt totally comfortable telling the White House “no” — as Zuckerberg himself admitted repeatedly on Joe Rogan, just weeks before telling Elon he was happy to silence people at the Trump White House’s request.
So the same political movement that treated government staffers sending cranky emails — emails that platforms felt perfectly free to ignore — as an existential constitutional crisis now sees nothing wrong with a formal State Department cable directing coordination with a specific privately-owned platform and military PSYOP. If the principle only matters when your political opponents are the ones in the White House, it was always just about weaponizing the systems of government for your own benefit.
Klonick puts the broader structural picture together:
The privatization of Twitter removed all traces of public accountability. The gutting of content moderation infrastructure removed operational resistance. The political alliance between the administration and the tech sector removed institutional resistance. And now a formal diplomatic cable removes the last pretense of arms-length separation between U.S. government messaging objectives and the platforms that carry them.
The legal questions that Murthy left unresolved—about when government pressure on private platforms crosses the constitutional line—will almost certainly be relitigated in this new context. But the more immediate reality is that the internet Americans and global audiences navigate is increasingly shaped not merely by the preferences of platform owners and advertisers, but by the strategic communication objectives of the U.S. government, implemented through platforms that have every financial and regulatory reason to cooperate.
This is the pattern we’ve watched unfold for years: wrap your power grab in the language of the thing you’re destroying. Call fact-checking “censorship.” Call attempts to expose foreign influence campaigns “the speech police.” Dismantle the institutions that actually did the thing you claim to value, then use the resulting vacuum to do exactly what you falsely accused your opponents of doing — only bigger, more openly, and with military coordination.
The sheer audacity of the sequencing is what makes all of this so infuriating. They spent years pointing at the Disinformation Governance Board and screaming “Ministry of Truth!” They shut down the Global Engagement Center while Rubio called it censorship. They destroyed the research infrastructure and the institutional knowledge that actually helped identify and counter foreign influence operations. And now, having cleared the field of anyone who might push back, they’re running their own influence operations through a platform with no independent oversight, no transparency mechanisms, and no institutional resistance — and they’re doing it openly, through formal diplomatic channels, in coordination with military psychological operations.
Klonick closes with the right question:
The question is no longer whether the government can use social media as a tool of statecraft. It already is. The question now is whether any institution—legal, normative, or structural—retains the capacity to check it.
Given that the people who claimed to care about checking government entanglement with social media are now the ones wielding it most aggressively — and spent years systematically destroying every institution that might have served as a check — don’t hold your breath.
Remember when Elon Musk told advertisers to “go fuck” themselves and then sued them for the crime of taking his advice? A federal judge has now dismissed that lawsuit — with prejudice — confirming what anyone with a passing familiarity with antitrust law already knew: companies deciding they don’t want their brands plastered next to extremist content aren’t engaged in an illegal conspiracy. They’re just making basic (probably pretty smart) business decisions.
When X Corp filed this case back in August of 2024, we walked through in great detail why the legal theory was fundamentally broken. Not broken in a “they pleaded it badly” kind of way, but broken in a “this theory does not describe an antitrust violation no matter how many drugs you’re taking or how convinced you are that the world owes you advertising dollars” kind of way. Judge Jane Boyle of the Northern District of Texas has now agreed, and the key section of her ruling is worth reading in full, because it says what we said at the outset: X has not suffered antitrust injury.
The court laid out the standard, quoting the Fifth Circuit, channeling the Supreme Court, on what counts as an antitrust injury:
The Supreme Court has distilled antitrust injury as being “injury of the type the antitrust laws were intended to prevent and that flows from that which makes defendants’ acts unlawful.” … “The antitrust laws … were enacted for ‘the protection of competition not competitors.'” … “Typical” antitrust injury thus “include[s] increased prices and decreased output.” … “This circuit has narrowly interpreted the meaning of antitrust injury, excluding from it the threat of decreased competition.” … “Loss from competition itself—that is, loss in the form of customers[] choosing the competitor’s goods and services over the plaintiff’s—does not constitute an antitrust injury.” … In short, the question underlying antitrust injury is whether consumers—not competitors—have been harmed.
Antitrust law protects competition, not competitors. X’s entire argument boiled down to: “advertisers chose to spend their money somewhere other than our platform, and that hurt us.” But that’s just… the market. That’s how markets work. Customers choosing not to buy from you because they don’t like what you’re selling has never been an antitrust violation, and the court made short work of explaining why.
Amusingly, the GOP — whose campaigns Musk has bankrolled extensively — spent decades pushing for exactly this narrow definition of antitrust injury, precisely to make cases like this harder to bring. Perhaps one of those politicians could have mentioned that before Elon filed.
But this case was never actually about winning an antitrust case. It was a warning shot at advertisers: give Elon your money or we’ll drag you through an expensive court process. A shakedown dressed up in legal filings. Indeed, after the lawsuit was filed, it was reported that part of X’s “sales” process was to threaten companies that they’d be added to the lawsuit if they didn’t advertise on the platform.
The court examined X’s theory from two different angles, and it failed both times. First, if the conspiracy was supposed to benefit competing social media platforms (like Pinterest, one of the defendants), X hadn’t alleged that any competitor was actually behind the boycott or pressuring advertisers to exclude X so the competitor could corner the market:
X has not alleged that the advertisers chose to do business with Pinterest—or any other social media company—as part of an agreement not to do business with X. Unlike the large hospital in Doctor’s Hospital, Pinterest is not alleged to be X’s competitor that wanted to exclude X from the market so that it could charge higher prices. In turn, unlike the network in Doctor’s Hospital, the advertisers did not decide to boycott X at Pinterest’s—or any other X competitor’s—behest to secure the competitor’s business. Instead, X alleges a conspiracy driven by advertisers not to further X-competitor social media companies’ interests but to pursue their own collective interests as to where they place their advertisements.
Second, if the conspiracy was supposed to eliminate competition at the advertiser level, the court found that GARM wasn’t acting as some kind of gatekeeper blocking X from accessing customers. It was just… advertisers deciding for themselves:
GARM is not an economic intermediary like the retailers in Eastern States. GARM did not buy advertising space from X to sell to advertisers nor did it, in such an arrangement, tell X not to sell directly to GARM’s customers. Rather, GARM was organized by advertisers and reflected their “avowed commitment to furthering [their] economic interests . . . as a group.” … Thus, if GARM is the obstacle to X reaching its advertiser-customers directly, then it is the equivalent of the advertiser-customers themselves deciding not to deal.
That’s the ballgame. Advertisers collectively deciding they don’t want to spend money on your platform — especially after you’ve told them to go fuck themselves and your platform has become a haven for content that damages their brands — just doesn’t state an antitrust claim. Imagine being so entitled that when the marketplace rejects your offering, you insist that it must be an antitrust attack on your rights to their money?
The court was so confident in this conclusion that it dismissed the case with prejudice and denied X the opportunity to replead, noting that the 165-paragraph complaint was already plenty detailed:
The 165-paragraph Second Amended Complaint contains no dearth of detail: if facts existed that GARM operated at an X competitor’s behest to put X out of business or that GARM advertisers sought to unfairly exclude competing advertisers from doing business, X would have pleaded those facts. The very nature of the alleged conspiracy does not state an antitrust claim, and the Court therefore has no qualm dismissing with prejudice.
When a court tells you the nature of your theory doesn’t work, that’s about as definitive a loss as you can get.
As we noted when the case was filed, the evidence X submitted in its own complaint actually undermined the case. One of X’s own exhibits showed GARM’s lead, Rob Rakowitz, explicitly telling an advertiser that GARM doesn’t make recommendations and that advertising decisions are “completely within the sphere of each member and subject to their own discretion.” Another email showed Rakowitz telling an advertiser asking about Twitter that “you may want to connect with Twitter directly to understand their progress on brand safety and make your own decisions.” This is the supposedly nefarious conspiracy that X spent years and untold legal fees litigating.
Separately, I have to mention the blatant forum shopping: X filed this case in the Wichita Falls Division of the Northern District of Texas, which was widely understood as a transparent attempt to land in front of Judge Reed O’Connor, known for partisan rulings and already presiding over Elon’s SLAPP suit against Media Matters. That didn’t work out — O’Connor recused himself, not because of his ownership of Tesla stock, but rather his ownership of some advertising firms who were defendants. The case got reassigned to Judge Boyle, and X still lost. In an ironic twist, X then tried to transfer the case to the Southern District of New York, only to have the court deny that motion because X couldn’t even prove they did business in that specific district. So X handpicked a forum, lost its judge, and then couldn’t escape to a different one. Great lawyering.
But the legal dismissal, satisfying as it is, doesn’t capture the most important part of what actually happened here. Because while the court correctly found that X suffered no antitrust injury, GARM itself suffered a very real injury: it was killed.
GARM shut down within days of the lawsuit being filed, following Rep. Jim Jordan’s misleading congressional investigation that painted the organization as some kind of anti-conservative censorship machine. Jordan’s pressure campaign, combined with the threat of expensive litigation from the world’s richest man, made it untenable for GARM to continue operating. The organization that existed to help advertisers make informed decisions about brand safety — a fundamentally expressive activity, protected by the First Amendment — was destroyed through government jawboning and litigation threats.
There was only one attack on free speech involved here and it came from Jim Jordan and Elon Musk, not GARM or its advertiser members.
X filed this lawsuit wrapped in the language of free speech. Former X CEO Linda Yaccarino literally wore a necklace that said “free speech” while announcing the case, claiming that advertisers not giving X money was somehow an attack on users’ ability to express themselves. The actual speech suppression ran the other direction entirely. A private organization exercising its speech rights to help its members make informed business decisions was bullied out of existence through a combination of congressional intimidation and frivolous litigation.
Jordan celebrated GARM’s dissolution as a victory for free speech — par for the course for the censorial MAGA GOP. A congressman used the weight of his office to pressure a private organization into shutting down, and called that free speech. Meanwhile, the lawsuit that was part of that same ecosystem of intimidation has now been found to have no legal merit whatsoever.
This is what actual jawboning looks like in practice. The lawsuit didn’t need to succeed to accomplish its goal. GARM is gone. The organization that facilitated conversations among advertisers about how to protect their brands has been silenced. The chilling effect on any future organization that might want to do something similar is obvious and intentional. Any industry group that tries to coordinate around brand safety now knows that it might face a billionaire-funded lawsuit and a congressional investigation for its trouble.
The court’s ruling is a vindication of basic antitrust law. But the more important point is about what the actual free speech dynamics were in this whole saga.
X can appeal, of course, and given that this falls within the Fifth Circuit, stranger things have happened. But the fundamental problem remains what it’s always been: the theory that advertisers owe you their business because you exist, and that organizing around brand safety is a criminal conspiracy, has never been a viable legal argument. The court said so plainly. Dismissed with prejudice. Nothing to fix, because the whole premise was broken from the start.
Full disclosure up front: I sit on the board of Bluesky. That said, I had absolutely no idea this lawsuit existed until recently. Which, honestly, tells you something about how much of a legal non-event it was. But the underlying story here—about the NFL treating social media the way it treats television broadcast rights—is worth digging into, because it reveals something deeply broken about how major sports leagues think about the internet.
The 2025-2026 NFL season just wrapped up, and along with it came a federal court ruling in a case called Brown v. NFL that most people missed entirely. Two football fans—one in Illinois, one in California—sued the NFL under the Sherman Act, claiming the league violated antitrust law by barring its teams from posting on Bluesky. The fans wanted to follow their teams—the Bears and the now-champion Seahawks—on the platform they actually use, rather than on Elon Musk’s X. The court dismissed the case for lack of standing, and honestly, that was probably the right legal outcome.
The fans couldn’t demonstrate a concrete injury—the information they wanted was still available, for free, on X. As the court put it, their grievance reduced to being “denied the ability to obtain real-time NFL team information on a private platform with which they are ideologically comfortable.” And “I don’t like Elon Musk” is not an antitrust injury. The Sherman Act targets conspiracies that restrain trade and harm competition—not content distribution preferences. You can’t force a private organization to distribute its content on the platform you like best, just as we’ve called out attempts to force social media platforms to carry content they don’t want to carry.
But the fact that the NFL is legally allowed to be this myopic doesn’t make it a smart business decision. You can be entirely within your rights and still be making a spectacularly bad call.
Since 2013, the NFL has had a “content partnership” with X (dating back to when it was the useful site known as Twitter). The deal lets X publish real-time highlights, and in return the league gets… money, presumably. As the court noted in its ruling:
Since 2013, the NFL and X (formerly Twitter, Inc.) have had a “content partnership.” It allows X to publish real-time highlights from football games, such as touchdowns. During the offseason, reporters post on X with news about team practices and other NFL-related topics, and fans on X discuss teams’ acquisitions of free agents and other roster changes. For example, during the NFL draft (the high-profile annual event in which teams select eligible players to join their rosters), X published more than one million posts concerning the NFL; these appeared on users’ screens more than 800 million times. The NFL has repeatedly renewed its partnership with X. Fans do not pay money to receive NFL news on X.
Fine. Lots of organizations have deals with social media platforms. But this just seems like self-sabotage: the NFL apparently used this partnership as justification to tell its own teams they couldn’t even exist on a competing platform. Multiple NFL teams—including the New England Patriots—had set up accounts on Bluesky, started posting, and were building audiences. And then the league office stepped in and told them to shut it all down.
From the ruling:
Initially, multiple NFL teams, including the New England Patriots, had accounts on Bluesky to communicate with fans….
As alleged, however, the NFL later instructed its member teams to delete their Bluesky accounts. But for this instruction, at least some NFL teams would use Bluesky. The Patriots’ vice president of content, Fred Kirsch, for example, has stated: “Whenever the league gives us the green light[,] we’ll get back on Bluesky.”
Yes, the (Super Bowl-losing) Patriots’ VP of content is publicly saying his team wants to be on Bluesky and is just waiting for the league to let them. This wasn’t a case of teams being uninterested. Teams saw the audience there, set up shop, and were actively communicating with fans—and the NFL made them stop.
As Front Office Sports reported at the time, the league specifically told the Patriots to take down their Bluesky account. The league apparently hasn’t even approved Threads—Meta’s X competitor—for team real-time updates either.
So the NFL has essentially decided that when it comes to the kind of real-time updates that fans actually care about, X is the only approved outlet. Everything else is locked out.
This is “broadcast-brain” thinking applied to the internet, and it’s spectacularly dumb.
The NFL is treating social media platforms the way it treats regional sports networks or its Sunday Ticket package: as exclusive territories to be carved up and sold to the highest bidder. In the television world, that model makes a certain kind of sense—there’s a limited amount of spectrum, a limited number of cable channels, and that scarcity creates value. But social media doesn’t work that way. There’s no scarcity. Posting an injury report on Bluesky doesn’t remove it from X. Cross-posting is literally free. The entire point of social media for a brand is to be everywhere your audience is.
And the audience, increasingly, is on Bluesky. As Mashable noted last year heading into the season, the NFL community on Bluesky had already hit a kind of critical mass:
You need the presence and regular posting of big names to legitimize a platform. It certainly helped that folks like Kimes and alargeportionof theNFLwritersat popular sports sites like The Ringer made Bluesky home. And last season it felt like Bluesky hit terminal velocity, where enough people joined that you could fully exit to the site for football content. And with the migration of the professionals, the shitposters naturally came along, too. Because that’s where the discussion was happening. There is genuine, easy-to-find, fun NFL talk on Bluesky with minimal interruptions from, say,weird adsorangry reply guysyou might find on X.
That’s a real community. A vibrant, engaged community of exactly the kind of hardcore football fans that the NFL should be desperate to cultivate. These are, as Mashable noted, the “ball knowers.” They’ve moved to Bluesky because, well, X kind of sucks now for following sports. As Mashable also noted:
Bluesky does have a leg-up in some areas — Elon Musk’s site recently has proven unreliable for NFL fans. Thesite crashed the morning free agency launched, which is one of the most important days for NFL social media. And the sports tab — which used to be an easy, fun way to follow games in the Twitter days —degraded into near uselessnessyears ago. And, in general, X has morphed with Musk’s image, which is focusedmore on AIand politics — not things like following football. Of course you can still follow the NFL on X, but it does involve wading through more junk than it used to. Bluesky offers an interesting alternative in that regard.
So the most engaged, most knowledgeable football community has moved to Bluesky. The teams themselves want to be on Bluesky. And the NFL’s response to all of this is… to ban its teams from showing up.
It’s the digital equivalent of a local blackout (something we’ve been calling out for well over a decade)—punishing your most dedicated fans because of some deal you cut with a middleman in an effort to create an artificial and unnecessary scarcity.
Meanwhile, the platform the NFL is propping up with this exclusivity arrangement is one where fans who tuned in for the Super Bowl halftime show got to watch a significant chunk of the X user base have a full-blown racist meltdown over Bad Bunny performing. The NFL specifically chose Bad Bunny to appeal to a broader, more global audience—and the audience that actually appreciated the choice? They were on Bluesky where there was an overwhelming wave of support for the performance. The league is betting its real-time presence on the platform where its expansion strategy gets shouted down, while blocking teams from the one where those new fans are actually showing up.
This kind of control-freakery from the NFL shouldn’t surprise anyone who has followed the league’s behavior over the years. This is the same organization that has spent decades aggressively lying to bars, restaurants, and small businesses about the scope of its “Super Bowl” trademarks, sending threatening letters suggesting you can’t even say the words “Super Bowl” in an ad without a license—something that has never actually been true.
The NFL’s institutional DNA is “control equals value,” and they apply that logic to everything, from what a church can call its viewing party to which social media apps their teams are permitted to use.
The problem is that control-based thinking only works when you actually can control the ecosystem. You can (sort of) control which networks broadcast your games. You can control which streaming service gets Sunday Ticket. You cannot control where fans choose to talk about football on the internet. The conversation is going to happen whether the NFL’s official accounts are there or not. The only question is whether the league’s teams get to participate in it.
Any organization whose core business depends on fan engagement should be finding fans where they are, not herding them onto a single platform because you cut an exclusivity deal. Especially when that platform is increasingly known for being a hellscape of AI slop, political rage, and engagement-bait, while the platform you’re blocking your teams from is the one where people are actually talking about your product with genuine enthusiasm.
The NFL generates billions in revenue. And yet, when it comes to social media strategy, it’s stuck in a 2005 mindset. That’s not how any of this works anymore.
Someone at NFL HQ needs to understand that when your most passionate fans have moved to a new platform and your own teams are begging for permission to follow them there, the smart play is to let them go.
