The administration offered this excuse for its lack of compliance with its own promises:
Whiletalking to reporters Tuesday, White House Border Czar Tom Homan said “hundreds” of cameras were purchased and sent to Minnesota following the shootings of Renee Good and Alex Pretti.
“But there wasn’t enough cameras to outfit every ICE agent, I was waiting up for more money,” Homan said.
MORE MONEY? That’s the excuse?!? Literally unbelievable. The “Big Beautiful Bill” had been passed monthsbefore the killings and nearly a year before Homan turned out his pockets to do his imitation of the “POOR TAX” Monopoly card.
The legislation makes U.S Immigration and Customs and Enforcement the largest federal law enforcement agency, giving it $45 billion for building new detention centers in addition to $14 billion for deportation operations. It also includes $3.5 billion for reimbursements to state and local governments for costs related to immigration-related enforcement and detention.
Having thrust its way back into the headlines in the worst way possible, DHS is finally setting a deadline for a full rollout of body cameras:
US Immigration and Customs Enforcement said Saturday it expects every officer and agent in the field to be equipped with a body camera by the end of August, a rapid expansion that could provide an unprecedented visual record of the federal government’s immigration enforcement operations.
Good news? Well, maybe. I mean, it’s been known for years that body cameras do more for law enforcement officers than they do for things like accountability or transparency. Pinning cameras to ICE officers probably isn’t going to alter that reality… especially when ICE has made it clear from the outset that it’s rarely, if ever, going to make body camera footage public.
[T]he agency’s body-worn camera policy could mean taxpayers who are pitching in tens of millions of dollars to pay for the cameras may be kept in the dark about some high-profile confrontations, at least in their immediate aftermath.
The policy says ICE will promptly release video of shootings and other encounters in which its agents cause death or serious injury only after determining “it is in the best interests of the agency” to do so. That gives its director discretion to release recordings that reflect well on the agency’s image while working to keep others secret, experts on body camera policies said.
This policy dates back to the first months of Trump’s second presidency. Issued on February 18, 2025 (and updated a day later), it gives ICE considerable discretionary control over body cam footage releases, even as it pretends ICE might release recordings of injuries or deaths “in custody” within 72 hours of their occurrence. By adding a bunch of asterisks to something that’s supposed to look like a release-by-default directive, ICE retains the power to block releases of footage deemed to be of public interest if it decides its own interests are more important.
ICE has responded to this reporting of things its body cam footage actually says by claiming what’s been reported isn’t actually what the policy will allow the agency to do. Here’s acting ICE director David Venturella with the spin:
“ICE policy permits expedited release when appropriate — and clarifies that it’s not appropriate to release footage when doing so could compromise investigations or privacy,” Venturella said. “This approach mirrors other federal law enforcement agencies’ policies and is fully consistent with the Freedom of Information Act and the Privacy Act.”
He continued: “At the same time, we’re committed to transparency and accountability, and we’re on track to equip every officer and agent in the field with body-worn cameras by the end of August, ahead of our original timeline.”
Venturella is affirming the same reporting he’s disagreeing with. One assumes every ICE shooting or confrontation resulting in severe injury will be “investigated.” Whether these investigations will be sincere and/or objective hardly matters. All that matters is that ICE can say the incident is still under investigation and refuse to release footage until the investigation has been concluded. There is no statutory time limit on internal investigations, which means that ICE can withhold this footage indefinitely. So, the reporting Venturella claims is false is actually just an accurate portrayal of the ICE’s body cam footage policy — a portrayal that’s been stripped of the legalese and dissembling ICE would prefer remain in place.
ICE officers are getting cameras… allegedly. But it looks like the general public will be unable to verify this rollout via ICE releases or public records requests. And officers who kill or main American residents and citizens will not only be allowed to review their recordings before filing reports/making statements to investigators (another feature/bug of the February 2025 policy), but are pretty much guaranteed their potentially damning recordings will be buried by the agency that employs them.
I spoke a little about this on last week’s Ctrl-Alt-Speech, but now that Anthropic has come out with more details about how its text watermarking works, the debate has shifted into overdrive. Many people are upset about it, even though it appears that Google has already been doing something similar with the output from Gemini. I think that people are right to be upset, but for the wrong reasons, aimed at the wrong target.
The real problem is that the EU’s AI Act is aimed at a threat that never really materialized, and the end result will fall hardest on the people who get the most genuine benefit from these tools. Also, it doesn’t help that Anthropic chose to comply with the law in a manner that looks broader than what the law requires — though it did so for reasons that are more understandable than some of its critics suggest. Either way, though, the costs will fall most heavily on people who are using the tech properly.
Let’s take a few steps back first. If Anthropic’s explanation of how they watermark text isn’t clear enough for you, I think this explanation of how text watermarking works is much better. The fundamental thing to understand about generative AI is that it’s always trying to generate the next token, and it does so probabilistically, not deterministically, meaning that each time you’ll get something slightly different. The results already have biases in them (that’s part of the weights part of LLMs), but the companies can deliberately bias them in a manner that the tool is ever so slightly more likely to choose certain words based on a prompt than without that bias.
Think of it in the way that most random generators are not, in fact, “random.” With a little bit of effort, people can often figure out the “bias” of a random number generator, giving them an advantage in determining what number will be generated. Here, it’s the same sort of thing, but the “bias” is impacting the randomness of which word the tool will choose next.
With a long enough text, and a key regarding the bias, you can look at the text and see that enough of the very slight changes match the “watermark” bias, as to suggest the text was likely generated by a model carrying that watermark. Such a system is hardly foolproof, but it can absolutely call out text likely generated with that specific bias. Editing text after the fact may or may not get rid of the watermark, depending on whether or not the edits remove/change enough of the “biased” words.
Many of the people who are upset are so because they think they won’t be able to cheat any more, and… I don’t care one bit about them. There is some concern that the biasing will make you choose worse words, but I’m also not all that concerned about that. As far as I’m concerned, if you’re using the AI to write for you, in some cases you’re already having it choose words, and it should be on you to know when and how to choose better words. The stronger version of that objection is about how the watermark will apply when merely editing content, in which case the watermark bias may nudge your prose towards specific synonyms. But even there, that seems much more like an argument to use the tools differently, not as a fully damning issue.
My larger concern is in how this will almost certainly be used to simply attack and denigrate people who actually are using the technology in a reasonable manner, but will be falsely accused of “cheating.” Just this week I wrote about the ways in which I use AI tools to help with my work, not as a tool for writing, but helping me review and edit stories. In that article, I mentioned the REAL Rating site, which offers a one-to-five scale to make clear that not all AI use is the same.
But a watermark reduces all of that to a simple binary: “did this use AI at all.”
And I worry about the fallout from that.
Which brings us back to where all this came from. As LLM tools started becoming popular, the EU rushed out its EU AI Act, making it clear that they didn’t want to be slow to regulate, in the way that they (falsely) feel they should have regulated social media much earlier. Part of the early concerns with AI was that it would be used for “deepfakes” to fool people. As we predicted, that fear has been largely overhyped. There are still concerns that it could become a problem, but to date it really has not been. Most AI-modified content has been rightly called out as such. Indeed, the biggest thing with deepfakes tends to be people who are legitimately caught on video behaving badly blaming deepfakes (usually unsuccessfully) for their actual actions.
But because that was one of the biggest initial fears, the EU included a provision demanding “transparency of AI-generated content.” Providers are now required to mark LLM-generated content so that someone can identify it as such. And if deepfakes are a problem you’re trying to defeat, that could make some amount of sense.
However, deepfakes aren’t really much of a problem.
What we do have is a large and understandably frustrated group of people who are hostile to any AI use at all. And now they’ve been handed a detector with which to accuse anyone of being an AI user. So instead of using AI detectors to prevent people from being fooled, the tech is now being turned around to accuse people of using the tools in any way, shape or form. And that seems like a problem.
To be fair, the Code of Practice does not demand transparency for text that a model merely cleaned up, translated, or spellchecked. Anthropic still went ahead and watermarked all of it anyway. You could argue that this is a form of malicious compliance, but the reality is that it’s a bit more complex than that. Adding in this feature just for fully generated text would require separating out different kinds of text generation, which may also confuse things. Also, for most (but not all) merely “processed” text, there’s a lower likelihood of the watermark showing anyway, since identifying the watermark requires a decently long string of generated words.
Along those lines, Anthropic admits that the watermark is less likely to show up in software code, since code usually needs to be more exact. There are just fewer interchangeable ways to write a working function, which means there are fewer opportunities to embed the kind of word choice randomness to make the watermark effective.
The company also says that it has no “durable” way to limit the watermark such that it only impacts EU users covered by the EU’s law, which is why it’s rolled the feature out globally. I have a little difficulty believing that. Geofencing websites is pretty common practice to deal with geographic regulatory restrictions, such that even if it’s not perfect, it’s not clear why this needs to be rolled out globally. Indeed, a company clever enough to figure out how to do a generated text watermark can also figure out how to do some basic geoblocking.
Back to the potential fallout of all this: First, let’s acknowledge that plenty of AI generated content does, in fact, suck. And I absolutely get the same instinctual negative reaction many people get when I see something that is clearly AI generated. It feels lazy and often annoying.
But not every use case is the same, and plenty of people reach for these tools for reasons that have nothing to do with laziness. For example, non-native speakers find these tools genuinely helpful in communicating more clearly with native speakers.
But now, attempts to use these tools to be a better communicator, may be dismissed as “fake” or “AI generated.” This is already a concern. A (potentially outdated) study from a few years ago found that non-native English speakers were more regularly accused of using AI tools. The same goes for Black students, who were more likely to be accused of using AI tools.
Yet, the setup of an AI watermarking tool is that it gives you a simple binary “yes” or “no.” And as much as Anthropic says that finding its watermark simply suggests that its tools were used, rather than making it conclusive, we all know that most people will use it as a sign of proof.
On top of that, the more sophisticated users — generally those with more knowledge and understanding — will likely find it relatively easy to use tools that will effectively remove the watermark. In fact, such tools already exist. That same site I linked to above with the clearest explanation of how the watermarking works, also hosts a tool built to strip the signal out. So the people who actually get dinged by this system will be the less sophisticated, less resourced users — exactly the ones with the most innocent reasons for using the tool in the first place.
And the impact there can be massive. There are already studies that suggest the mere labeling of something as having been mediated by AI causes people to trust it less, even if the underlying content is the same. There’s even one study that found that truthful information, if it has an “AI disclosure” on it, is often deemed as false, even if the actual information is true. That seems quite unhelpful!
So the EU’s effort to ward off the potential threat of deepfakes that hasn’t materialized, has created a very real problem: now designating people who are using the tech in useful, helpful, ways as “cheaters” whose content can’t be trusted. And I’m not the only one noticing this. As I was finishing up writing this, John Gruber pointed me to a blog post written by James Padolsey (the creator of the DeClaude tool above, and the writer of the excellent explanation of watermarks), with the fantastic title, “Anthropic’s weak watermarks appease a weak law” in which he writes:
That is to say: this rule risks penalising people who are already less able to produce conventional prose unaided for using technological means to make their lives easier. The Act exempts standard editing, but many legitimate assistive uses require more substantial rewriting while leaving the ideas, judgment and responsibility with the human.The same thought that led to this law could have applied to calculators at the time of their inception, had their outputs revealed themselves through artefacts. Thankfully, a sum borne of the brain is treated no differently from one produced by a calculator. Likewise with spellcheckers. To make assistance suspect only once the tool becomes capable enough to compose a whole sentence is not a principled boundary. It is a moral premium placed on difficulty itself.
Anthropic has nevertheless chosen a blanket, model-level implementation that appears broader than the law’s minimum requirement. That may be convenient compliance engineering, but it discards distinctions the law expressly attempted to preserve.The result is a signal broad enough to implicate harmless and assistive use, yet fragile enough to be removed by a motivated person through substantial recomposition. It risks concentrating suspicion on ordinary and assistive users while remaining weakest against deliberate deception.
Indeed. So we have a mistargeted law, badly drafted to go after a theoretical problem that hasn’t proven to be real, combined with penalties harsh enough that companies like Anthropic comply in the broadest, bluntest way possible — and all of it dropped into a world that has increasingly decided that AI use is a binary good-vs-evil question, and where no nuance is allowed.
The watermark won’t actually catch the people it’s ostensibly aimed at — those seeking to deceive people. Those people will likely be sophisticated enough to remove any such watermark and walk away clean. However, it will likely catch the people who had every right to use the tool in the first place: the non-native speaker who ran their draft through Claude to sound more fluent, the assistive-tech user who needed help composing a sentence, or the writer who wants an extra level of review on any text they’ve written. They’re the ones who will wear the scarlet letter. The EU wanted to look tough on AI and Anthropic wanted to look compliant. But neither of them has to answer to the the disabled user who used the tech to help them communicate, who now gets accused of lying and cheating.
Sure, just about anyone can point and shoot with a camera, but only a true expert can snap a photo that’s worth sharing with the world. From exposure to shutter speed and everything in between, this award-winning training walks you through the nuts and bolts behind capturing truly stunning photos. iPhotography is the leading online photography choice of Fortune 500 companies, schools, training institutes, and aspiring photographers worldwide, and it’s easy to see why. Each of this course’s 18 modules are interactive, up-to-date, intuitive, stimulating, and designed to be fun, and feature personal mentorship from award-winning photographers along the way, taking you from zero to hero with your camera in no time. It’s on sale for $40.
Note: The Techdirt Deals Store is powered and curated by StackSocial. A portion of all sales from Techdirt Deals helps support Techdirt. The products featured do not reflect endorsements by our editorial team.
Flock’s ALPR (automated license plate reader) tech spread from HOAs to police departments at an alarming rate of speed. What was once a toy for the over-privileged soon became the go-to source of license plate images. Flock became the Ring of ALPRs, to mix a metaphor. And that’s when the bad news began to roll in.
Everything happened pretty much all at once. Cops were revealed to be using this tech to track people who simply wanted to seek legal abortion options in other states. Federal officers were revealed to be leaning heavily on local law enforcement to generate Flock ALPR searches federal officers weren’t legally allowed to perform on their own. And once the spread of Flock cameras reached an inflection point, cops did what cops have always done:
When blessed with persistent tracking tech, far too many officers tend to run searches targeting ex-wives, ex-girlfriends, their new paramours, and anyone else an officer might have a less-than-professional reason to be tracking.
Flock has felt the heat. It is now making extremely belated changes that might limit abuse in the future, but only if cop shops don’t choose to opt out of the default settings. Meanwhile, cities all over the nation are ditching Flock tech. And they’re finding out it’s almost impossible to do because Flock seems to prefer activated cameras to complying with the desires expressed by their now-former customers.
Flock is struggling to defend its contribution to easily abused surveillance. And cops aren’t doing themselves any favors by continuing to abuse this surveillance tech. Because no one asked him to do it, Atlantic contributor Charles Lehman has decided that now is the best time to defend Flock.
Given that background, it comes as no surprise that Lehman’s defense of Flock is as abhorrent as it is abysmally stupid. Let’s go to the leadoff, which suggests that there’s not enough data to support arguments against the efficacy of Flock’s ALPRs — an argument that deliberately chooses not to engage the data that has already been collected in this country.
Opening with anecdotal “evidence” provided by a single podcast guest, Lehman rolls into a mess of his own making:
Although research on Flock’s effectiveness is still in the making, ALPRs are likely a valuable tool for any modern police force. Privacy concerns can be best addressed through smart regulation, not bans. Indeed, in our surveillance-saturated society, police cameras can help make the criminal-justice system both fairer and less punitive—if we let them.
Lehman opens up with an assumption he can’t back with data (“likely”). Then he heads directly into claiming mass surveillance tech can be brought to check by “smart regulation.” Finally, he makes the literally unbelievable claim that adding more surveillance tech will make policing less biased and more forgiving — something that has never been the case no matter how much tech cop shops buy. If anything, adding surveillance tech tends to amplify these existing problems by allowing cops to target whoever they want while feeding compromised data to systems that are “trained” to output garbage if their only inputs are garbage.
Lehman is a rube: a guy who hears that this one time Flock cameras caught a guy in a stolen car and has extrapolated that singular event to signify a wholesale improvement in public safety. He’s no different from the people who see the 1-in-a-thousand “good guy with a gun” takedown of a mass shooter and declare gun control to be a public harm.
But let’s allow Lehman to speak for himself… at least as far as he’s capable of doing so:
The case for Flock is simple. Cameras help police catch criminals. That reduces crime through two channels: incapacitating offenders who are caught and convicted, and increasing the certainty of apprehension, which theoretically should deter criminals from offending in the first place. But is it true in practice? Although the research base is still developing, the answer is “probably.”
When your closing argument is “probably,” you’re basically just overstating the probability of “maybe?” and hoping no one will notice the difference. Dumping a “theoretically” between two statements you want to connect by bridging them with a term that scientifically demonstrates uncertainty absolutely subtracts from the conclusion Lehman hopes his fellow rubes will derive from this poorly-conceived construct.
You may think this argument in favor of a massive network of Flock cameras couldn’t get any stupider. But that may be because you’re hoping Lehman himself couldn’t get any stupider because you’re actually starting to feel bad for him — like he’s a substitute gym teacher who suddenly got asked to fill in for an absent sociology professor.
Don’t. He wanted this published and the Atlantic agreed to it. Let’s just let him wallow in his amazing arguments in favor of his “Flock is good actually” theory, especially when he decides the best proof that surveillance tech works can be found in countries where human rights and civil liberties are barely an afterthought:
An analysis exploiting the quasi-random distribution of cameras in Medellin, Colombia, found that they reduced crime by about 19 percent. Another paper showed a 25 percent reduction in crime in subway stations in Stockholm, Sweden, after cameras were installed in them. A third found a 20 percent reduction in areas of Montevideo, Uruguay, where cameras were introduced. And a fourth estimates that nearly tripling cameras per capita in China reduced crime by about 10 percent.
Medellin’s violent crime rate has been dropping for years. The recent introduction of cameras hasn’t appreciably changed anything about this downward trend.
Citing a study focused on Stockholm is just inadvertently funny, because no one has ever thought of Stockholm as a criminal hellhole in need of maximal surveillance. While crime rates have been trending upward, some of that is due to law enforcement’s positive relationship with the people they serve — something that tends to result in more crimes being reported because the public actually feels law enforcement cares and will try to do something about it.
(And because I’m going to be far more fair than Lehman when dealing with this issue, I would assume some of our historical lows in crime rates are due to US citizens preferring not to interact with US law enforcement if at all possible, especially when they know some criminal acts will be ignored and that far too many cops will use a criminal investigation excuse to engage in warrantless searches, subject victims to harassment, or do the bare minimum needed to pencil whip an investigation checklist.)
But let’s really focus on the last sentence: here’s a person arguing in favor of persistent surveillance by citing supposed success in China. Let that sink in for a bit. Even ignoring the fact that Lehman thinks something that happened in a totalitarian nation supports his argument for increased surveillance of US people, we’re still left with the even more unseemly aspects of this citation in favor of Lehman’s pro-Flock theories: that crime reduction stats provided by a nation-state that completely controls the narrative are trustworthy. Good lord, man. Have some self-respect. Because if this is what you think is trustworthy data, you’re going to do nothing but spin in whatever direction Flock Safety PR reps or local law enforcement officials tell you to spin.
Lehman tries to temper that by saying he knows some people “will blanch” at using data from China to support ALPR cameras. And while he claims to recognize a trade-off is being made (often without our consent) to sacrifice privacy for law enforcement “efficiency,” he uses the term “security” to disguise the fact that this is just about cops wanting to get everything they want as quickly and easily as possible, no matter what harms it may pose to US residents and their rights.
After all of this fumbling towards a justification for increased surveillance, Lehman attempts to tie things up with a paragraph that isn’t actually supported by his arguments, his cherry-picked data, or his momentary asides in which he claims he’s respectful of civil rights and liberties:
Flock cameras are an easy target for a populist backlash. But although some of the concerns are reasonable, a panic about a “slave state” shouldn’t determine public policy. Intelligently regulated, ALPRs can be one among many tools in the toolbox of smarter—and therefore less severe—police forces. Dumping the cameras would just be foolish.
This is all bullshit. First, Lehman tries to diminish any rejection of mass surveillance systems and opposition to Flock and its tactics/services as “populist.” That’s pretty rich, coming from a fabulist who thinks he can add 2+2 and get 5 by pointing to data that doesn’t actually say what he thinks it says (and data that only says what the Chinese government wants it to say). Then he conjures up the theory of “intelligent regulation” before making it clear that any regulation he would consider to be “intelligent” would not result in the ditching of Flock ALPR systems. According to Lehman, “intelligent regulation” is limited to regulation he personally agrees with. Anything else is just “populism.”
To sum up: Lehman says persistent surveillance is right and everyone else is wrong. So long as it results in a few more arrests, the public should consider itself lucky to involuntarily be relieved of its constitutional protections.
Back in 2014 you might remember there was a giant, two-year, manufactured news cycle where Republicans pretended to be outraged that the then Obama White House had openly supported the FCC’s efforts to impose some modest net neutrality rules (designed to protect consumers and competition from telecom monopoly abuse of power).
At the time, major press outlets endlessly parroted Republican claims that the president’s perfectly legal public support of FCC policy was somehow underhanded, illegal, or applied undue pressure on an independent agency. Right wing friendly “think tanks” like the Hudson Institute suffered endless embolisms over the purported impropriety:
“The unwelcome assault on the independence of the FCC is a threat not merely to that agency. It is a threat to our entire system of government based on the rule of law. It is also a threat to the Internet. Where there is no rule of law, the Internet does not flourish.”
The Republican Committee on Homeland Security and Governmental Affairs even launched a two-year investigation culminating in a bogus report falsely claiming that there had been all sorts of nefarious intermingling between the White House and FCC. Republicans, I’ll repeat, were absolutely outraged that the White House would even open its mouth to comment on policy the agency was already pursuing.
Twelve years later and not only is net neutrality dead as a doornail thanks to corrupt collaboration between Republicans and telecom monopolies, but the FCC has taken up permanent residence inside the White House’s colon, making any claim of FCC independence laughable under Trumpism.
Or at least that’s the takeaway of this new report by the Guardian, which obtained documents via a Freedom of Information Act request and a lawsuit filed by the advocacy group Democracy Forward.
“Between 12 March 2025 and 11 February 2026, Carr attended at least eight meetings at the White House, according to the records, and had at least three scheduled calls with administration officials, including two conversations with chief of staff Susie Wiles. He also spoke by phone with administration officials such as controversial health secretary Robert F Kennedy Jr.
In addition, the files provide insight into Carr’s media strategy, including communications with producers at Fox News, and make public previously undisclosed meetings he has held with media titans such as Fox Corp chief executive Lachlan Murdoch and Sinclair Broadcast Group chair David Smith.”
That Carr is ignoring any pretense of agency independence, coordinating messaging with right wing agitprop factories like Fox News and Sinclair, and disproportionately meeting with right-wing friendly companies to their exclusive benefit shouldn’t surprise anybody and isn’t illegal (in a country with dead or dying corruption, lobbying, and influence-peddling laws), but it is a notable new high watermark for Republican hypocrisy. Twelve years ago, I’ll repeat, Republicans (including Carr, who had previously worked as an attorney at the agency), insisted the president even opening his mouth on policy was somehow illegal and an outrageous erosion of FCC independence.
The FCC is, to be clear, supposed to be independent from White House influence and meaningfully overseen by Congress. Though you might recall that last year in congressional testimony Carr raised eyebrows by insisting the agency is no longer independent — even going so far as to delete the word from the agency website.
Having regulators that are marionettes for the whims of your autocratic leadership is obviously corrupt and anti-democratic. Now it’s impossible to determine where Donald Trump’s incoherent whims begin and the FCC’s purported autonomy ends as Brendan Carr harasses media companies critical of the president, and strips away what’s left of U.S. consumer protection guardrails down to the studs.
Ultimately, I suspect the closest Carr will ever get to meaningful punishment is a revolving door gig at an AT&T-funded think tank where he whittles away the rest of his years getting paid high six figures to perpetuate the ongoing destruction of whatever’s left of U.S. public interest guardrails.
The past 18 months have seen a significant number of health issues, perhaps crises, arise in America. Measles, cyclosporiasis, whooping cough, and more have all been in the news. RFK Jr. is either stroking his pet conspiracy theories alongside Trump, or else he’s completely out to lunch. In the meantime, Kennedy is also continuing to sow doubts about vaccines generally, while also trying to claim he’s not, which helped result in a Trump executive order to reduce vaccination schedule recommendations for children. It’s as though this administration saw before it a four alarm fire and asked out loud, “How can we make this worse?”
The Centers for Disease Control and Prevention on Monday released data on the vaccination rates of US kindergarteners in the 2025–2026 school year, revealing that rates have once again decreased from the previous school year.
Vaccination rates have been slipping since the 2019–2020 school year, when the US had obtained rates of around 95 percent, the threshold needed to keep outbreaks of vaccine-preventable diseases from spreading in communities. The latest data shows the US is moving further away from that target.
Nationally, only 92.4 percent of kindergartners were vaccinated with the measles, mumps, and rubella (MMR) vaccine in the 2025–2026 school year, down from 92.5 percent last year. Coverage with the DTaP vaccine, which covers diphtheria, tetanus, and acellular pertussis (whooping cough), similarly fell from 92.1 percent to 92.0 percent.
This is bad quite generally, but it’s not equally bad in every state. And it’s very easy to demonstrate that state and local policies when it comes to vaccine requirements for school and the availability for non-religious exemptions are having quite the effect. This data is partially from the last school year, for instance, but it does a nice job highlighting the differences between states. In California, where exemptions face relatively tight requirements, 95.7% of kindergartners had the MMR vaccinations. In Florida, where the DeSantis administration has attempted to end all vaccination requirements for public schools, 88.9% of kindergartners had their MMR vaccinations.
The messaging from state and local leaders matters. The same is true for federal leaders. Public and health policy matter. And when we fall below herd immunity vaccination levels, the consequences are, well, precisely what we’re seeing out in the American wild at the moment.
If the overall number of exemptions exceeds 5 percent in any given location, it means that the location will not be able to achieve the 95 percent coverage target. Currently, 24 states have exemption rates at or above 5 percent. That’s up from 14 states in the 2023–2024 school year.
The data lands as vaccine-preventable diseases continue to rise and spread in the US. The country is on the verge of losing its measles elimination status as cases have reached a 34-year high and are still ticking upward. Last week, President Trump and Kennedy continued their efforts to undermine childhood vaccines, announcing an executive order to reduce the number of recommended vaccinations. The order has no scientific basis, and at a White House signing event, Trump and Kennedy spread vaccine falsehoods, misinformation, and disinformation.
This will get worse before it gets better, in other words. And the only real question is how many people have to get sick and/or die before we correct course.
It was late March when Joe Teirab, the second-in-command at Minnesota’s U.S. attorney’s office, received an urgent email from Washington.
The federal government was scrambling to find criminal cases to back up President Donald Trump’s claims that illegal voting by noncitizens was tipping the scales in American elections. Agents from Homeland Security Investigations, a massive federal law enforcement agency, had been dispatched to work leads across the country, including hundreds in Minnesota.
Teirab was already under pressure. In an earlier missive, Nick Davis, a high-ranking Justice Department appointee helping to lead the election fraud crusade, had reminded him the cases were so high priority that Teirab and his staff couldn’t decline to move forward on them without express approval from agency higher-ups. On March 24, Davis demanded a status report — within hours.
Teirab, a former Marine and a Harvard Law graduate who’d run unsuccessfully for Congress as a Republican, responded with a blunt reality check.
“Bottom line up front,” he replied in an email reviewed by ProPublica. After subpoenaing records on about 130 people, only one had been referred for prosecution, his staff had told him. Agents had deluged local election offices with calls and demands for voting histories, demonstrating “a complete lack of understanding” of illegal voting investigations.
“The HSI task force has been disjointed and disorganized,” Teirab wrote. The entire process, he said, had been “dysfunctional.”
Since Trump regained the White House, his administration has launched a series of unprecedented initiatives to find and prosecute voting by noncitizens, which he’s long claimed, without evidence, is rampant.
He’s stepped up this push in recent weeks, saying in a nationally televised speech that the American election system was “so vulnerable that no one can possibly defend it.” To support that assertion, the Department of Homeland Security, HSI’s parent agency, released documents asserting it had found more than 250,000 noncitizens on voter rolls in just four states, all led by Democrats. The documents included no explanation of how that number was calculated.
It’s well known the administration has tasked HSI — a force established to combat drug cartels, terrorism and other cross-border criminal enterprises — with leading the campaign to find election fraud cases in the United States.
But an investigation by ProPublica reveals for the first time how the Trump administration came to harness HSI’s personnel, technology and sweeping legal authority in service of its election agenda — and how meager the results have been, despite the prodigious resources sunk into the effort.
According to interviews and internal emails reviewed by ProPublica, career staffers at the Justice Department warned that transferring voter rolls to HSI to enable it to search for noncitizen voters could violate federal privacy laws. Similarly, longtime HSI insiders cautioned that using the agency’s databases and tools to search these lists would yield mismatches and wildly inflated results.
The administration plowed forward anyway.
HSI’s involvement in the hunt for election fraud traces at least to summer 2025, when agency supervisors embraced a proposal from a midlevel agent who’d publicly echoed Trump’s claims about elections. He argued the agency’s powerful databases and tools could find noncitizens even on the voter lists that states make publicly available, from which the most confidential information has been redacted. Under Trump, the Justice Department had collected many public voter rolls as part of a controversial effort to compel states to turn over the unredacted versions.
Those searches went forward, ProPublica’s reporting shows, helping to generate tens of thousands of leads regarding illegal voting across the country. But when HSI agents were sent to investigate them, the results were similar to what Teirab reported from Minnesota.
Between January 2025 and May 2026, a ProPublica analysis of Justice Department data shows, fewer than 150 alleged noncitizen voters were referred for prosecution. Even fewer — 41 — were charged with voting illegally or other election-related crimes. (More than 150 million people voted in the most recent presidential election.)
In response to questions from ProPublica, administration officials insisted the effort spearheaded by HSI was producing meaningful results.
“President Trump is committed to ensuring that Americans have full confidence in the administration of elections, and that includes totally accurate and up-to-date voter rolls free of errors and unlawfully registered non-citizen voters,” Abigail Jackson, a White House spokesperson, said in response to ProPublica’s request for comment. “Noncitizens voting is a crime. Anyone breaking the law will be held accountable.”
A DHS spokesperson didn’t answer questions about why so few prosecutions have resulted from HSI’s work. The agency wouldn’t specify what tools or techniques HSI had used, but confirmed it had cross-referenced “publicly available data” from state voter rolls with information on “known illegal aliens” in its systems. “It’s not rocket science,” the spokesperson wrote of this initiative in a response to ProPublica’s questions. “It’s an easy step to secure our elections.”
Teirab and the U.S. attorney’s office in Minnesota declined to comment.
There’s an array of reasons why Trump’s campaign to document claims of widespread voter fraud hasn’t succeeded. Most obviously, it’s exceedingly rare, as countlessstudies and state audits have found. Noncitizens often get on voter rolls by accident or when government officials make errors. Last month, New Jersey disclosed that a mistake involving its Motor Vehicle Commission caused 6,600 noncitizens to be registered (fewer than 400 voted).
The administration’s critics say its hunt for noncitizen voters aligns with Trump’s attempts to seek more federal control over elections while stoking doubt and fear about the voting process. Since taking back the Oval Office, Trump has tried to impose new restrictions on voter registration, mail-in ballots and voting machines, though judges have shut down most of these efforts.
Former officials at DHS and the Justice Department called the move to involve HSI a further escalation and questioned the propriety of aiming the agency’s muscle and technology at individual cases of illegal voting.
“It’s one thing if you’re going after Pablo Escobar,” said Steve Bunnell, a former DHS general counsel and senior intelligence adviser who handled voting fraud prosecutions during more than a decade at the Justice Department. “It’s another thing if you’re going after some cleaning lady who’s been working in the United States for 20 years and taking care of her 80-year-old mother and taking her little kids to church.”
“The Swiss Army Knife of Federal Law Enforcement”
Trump’s enlistment of HSI in his election fight is no accident. Cobbled together in the aftermath of 9/11, the agency has 7,100 armed, highly trained agents and another 800 criminal analysts, as well as access to troves of confidential data about hundreds of millions of Americans.
Though part of Immigration and Customs Enforcement, the agency has historically stayed out of immigration cases. During the first Trump administration, leaders of 19 HSI regional offices — virtually its entire top field hierarchy — signed a letter calling for HSI to become a standalone agency, arguing that ICE’s deportation work dissuaded people from cooperating with its investigations. But when Trump returned to the White House, it swiftly became clear that the agency would play a central role in investigations related to the president’s twin obsessions, illegal immigration and noncitizen voting.
Trump picked his field general for repurposing HSI even before taking the oath of office. In December 2024, he named Anthony Salisbury, a 50-year-old career agent who’d run the agency’s Miami office, as a deputy homeland security adviser, reporting to White House policy chief Stephen Miller. He was also given a dual appointment as head of HSI.
Salisbury was colorful: A mixed martial arts enthusiast, he once appeared at the agency’s headquarters with his face bruised and eye blackened from a recent bout. He’d sometimes entertain colleagues by pulling out a bridge that covers his missing front teeth.
To agency veterans, however, he was forever linked to a 2011 operation he’d overseen in Mexico in which one agent was killed and another wounded in a highway ambush by a drug cartel. An agency review submitted to Trump in April 2020 concluded that errors by HSI supervisors, including Salisbury, contributed to the outcome; it urged the administration to consider disciplinary action. None was taken against Salisbury, however.
Salisbury did not respond to requests for comment from ProPublica. A White House official called him “a critical member of the Trump administration” who was “cleared” in every review of the Mexico incident and was subsequently “promoted six times under multiple administrations.”
Once elevated to his new posts, Salisbury took charge of carrying out the second Trump administration’s agenda for HSI, personally issuing staffing directives, reassignments and promotions, current and former agency officials told ProPublica. (A number of them spoke on condition of anonymity out of fear of retribution.)
Starting in early 2025, Salisbury oversaw the unprecedented reassignment of more than 6,000 agents to immigration enforcement, diverting most of the agency from its normal duties. He then also pointed HSI at noncitizen voting, presiding over multiple meetings with officials at DOJ and Homeland Security focused on election fraud, according to emails reviewed by ProPublica and agency supervisors who worked with him.
“Stephen Miller has an HSI deputy for a reason,” said Eric Balliet, a high-level HSI manager who retired in 2024 after 23 years with the agency. (Like many HSI veterans, Balliet has remained in close contact with former colleagues.) “Salisbury is going to salute and execute, and he is going to make sure that from the HSI side, they fall in line, and there’s going to be no resistance or pushback. HSI has been turned into the Swiss Army knife of federal law enforcement.”
In early July 2025, Frank Quiñones, an HSI special agent who’d worked under Salisbury in Florida, approached leaders at the HSI Innovation Lab with an idea for using the agency’s technology to find noncitizen voters on state voter rolls, sources at the agency told ProPublica. Quiñones had been transferred to Washington to oversee a unit that had previously handled cases involving the theft of government benefits but that had been enlisted into voting investigations.
The lab, housed in an unmarked office in a D.C. suburb, had access to the government’s most sensitive databases, from suspicious activity reports to arrest records. Staffed by a combination of HSI experts and outside consultants, it developed software tools to comb the data for information that could help agents pursue criminal suspects.
Quiñones was a true believer in Trump’s claims about election fraud: In multiple Facebookpostsandreposts, ProPublica found, he promoted claims that the 2020 presidential vote had been stolen. At a meeting with the Innovation Lab’s overseers, according to sources who worked in the lab, he pitched using the lab’s technology to identify illegal voters — even though the agency lacked voter rolls that included identifiers such as partial Social Security numbers. (Quiñones did not respond to ProPublica requests for comment.)
For proof of concept, Quiñones proposed using a February 2021 public voter list he’d obtained for New Jersey and running it through HSI’s databases. The lab staff viewed his idea as “a little insane,” one recalled — both unreliable and improper. Since the public rolls don’t include voters’ unique identifiers, people at the lab also knew linking them to HSI’s data would produce mismatches — what the staffer called an “ungodly” number of false positives. The lab team also worried about violating longstanding safeguards limiting use of private citizen data. They “didn’t want to touch this,” the staffer added. Quiñones defended his idea, repeatedly declaring: “The president wants this!”
Tom Hodge, an HSI data analyst at the meeting, proposed running the voter information through the Athena Toolbox, an analytical platform developed by the Sandia National Laboratories that had access to all of HSI’s databases. (Hodge did not respond to ProPublica requests for comment. Sandia referred ProPublica’s questions to the facilities’ parent agencies, including the Department of Energy, which did not respond to our requests.)
Hodge and the Sandia team spent the next six weeks on the project, reporting they’d found large numbers of noncitizens on Quiñones’ 2021 New Jersey list. According to one former HSI official, they said they’d found “5,000 high-confidence illegal voters” in just a single New Jersey county. (Officials at the New Jersey secretary of state’s office, which oversees the state’s voter rolls, declined to comment on the claim. Beth Thompson, head of a group for local New Jersey election officials, called the number impossibly high, even accounting for the mistake disclosed by the state.)
The apparent breakthrough couldn’t have come at a better time.
The Trump administration’s other efforts to identify noncitizens on state voter rolls were running into a variety of roadblocks.
The administration had hoped to persuade states to check their voters’ citizenship status using a system called the Systematic Alien Verification for Entitlements, or SAVE, but many didn’t. Most states also refused the Justice Department’s demands to turn over their unredacted voter rolls, including partial Social Security numbers, which thwarted the agency’s plan to run them through SAVE. Courts have stymied the DOJ’s attempts to sue for the records, citing the Constitution, which gives primary control over elections to the states.
In an Aug. 21 email to a half dozen administration officials, Quiñones touted the solution he’d brought to the HSI Innovation Lab: using HSI’s platform to search voter rolls. The results, Quiñones noted, could fuel both criminal and immigration investigations.
By the end of the month, those in top election-related roles at DHS and the Justice Department were actively promoting what they dubbed the “HSI Tool” as a preferred alternative to SAVE, according to people familiar with the matter. Among them was Heather Honey, the prominent election denier appointed as a senior counselor at DHS, who is helping lead the government’s efforts to identify noncitizens on voter rolls. (Honey did not respond to ProPublica’s requests for comment. In an email responding to questions to her, DHS said Honey’s “expertise in election administration” was “invaluable to the Department’s efforts to protect critical infrastructure.”)
The White House, too, got behind the idea. In a November email, J. Brian Sikma, special assistant to the president, excitedly noted that the New Jersey trial cross-checking voter rolls with DHS data appeared to have identified a “very significant number” of “potential non-citizens.” (Sikma didn’t respond to a request for comment.)
The DOJ, Sikma reported, had already obtained public voter rolls for many other states. He included a list of them, adding: “It is of paramount importance that these also be reviewed expeditiously.”
Running “Roughshod” Over Privacy
As the push to use HSI’s tech to search voter information gained momentum, a new problem emerged.
Career attorneys at the Justice Department, including specialists on privacy law, raised concerns that transferring voter rolls wholesale from the agency to DHS might not be legal, according to internal emails reviewed by ProPublica and interviews with several former officials.
The voter rolls the DOJ had collected contained sensitive information on millions of Americans. Even the public versions, which political parties and candidates routinely obtain to target ads and messages, included voters’ addresses, birth dates and party affiliations. For the 16 states that have agreed to share their unredacted voter lists, the DOJ had citizens’ partial Social Security numbers or driver’s license information, too.
Federal laws, particularly the Privacy Act, dictate what data government agencies can collect, what it can be used for, how it can be shared, and how it must be protected. Agencies have to disclose their plans in advance, gathering public comment. A person’s information can’t be released or shared without their consent, subject to limited exceptions.
From the outset, the Trump administration’s effort to combine data across federal agencies has repeatedly drawn criticism from courts and whistleblowers for failing to adhere to restrictions meant to keep data private and secure. In June, for example, a judge prohibited the government from using SAVE for mass searches, ruling the administration had violated federal privacy laws by giving DHS access to Social Security data to enhance the tool. The administration has appealed that ruling.
In July 2025, as word got around that the DOJ might hand over voter data to DHS, a half dozen career attorneys who had been reassigned into the voting section voiced their unease to supervisors, according to two former DOJ lawyers. Their concerns were disregarded, prompting most to resign.
Instead, the DOJ’s office of legal counsel pushed forward with plans to share the public voter rolls, claiming a Privacy Act exception allowed HSI to receive such data without public notice as long as it was for law enforcement purposes. DHS just needed to submit a letter officially requesting the data for those purposes, an attorney in the legal counsel’s office explained, according to internal emails.
But this exception was meant to be used to get information on individual criminal suspects, not to gain mass access to data on people suspected of no wrongdoing, according to Nikhel Sus, chief counsel for Citizens for Responsibility and Ethics in Washington, which has sued the Trump administration on behalf of voter and pro-privacy groups, filing the case that eventually limited SAVE’s use.
Peter Winn, a 29-year Justice Department veteran who had served as the agency’s acting privacy chief for nearly a decade, drafted a letter to set out the data-sharing agreement between the DOJ and DHS, according to internal documents reviewed by ProPublica.
In a Feb. 12 email to two top DOJ officials, he said he’d rewritten an initial draft memo to “make the information sharing arrangement far easier to defend, if and when it is challenged in court.”
Rather than simply handing over the voter rolls, Winn proposed having DHS request “pertinent and relevant” information about specific people from the DOJ, then having technical staff at Justice and Homeland Security “coordinate” to fulfill the requests.
“Having slept on it,” Winn wrote, “I ended up deciding that our initial idea of sending a huge amount of raw unprocessed voter registration data to DHS, involved taking far too much unnecessary legal risk, given that 99% of the data would implicate the privacy and civil liberties of United States Citizens, and would risk adding to the false narrative in the establishment press of DOJ and DHS being unconcerned with the privacy and civil liberties of Americans.”
Winn attached his proposed letter, to be signed by Todd Lyons, then the acting director of ICE.
Winn’s suggested restrictions were quickly abandoned, however. A second draft of the letter, dated March 4, contained altered language providing that the voter information would be “transferred” to HSI, with one exception: Voters’ party affiliation would be redacted, if possible, it said.
The final letter, signed by Lyons and sent to the DOJ just two days later, abandoned even that limitation. Just three paragraphs long, it permitted HSI to use the voter rolls to pursue any “appropriate investigation of potential violations of federal election law.”
In response to questions from ProPublica, Winn said “I can’t really comment on internal drafts of correspondence.”
A Justice Department spokesperson defended the government’s actions, saying “no one should oppose intergovernmental data sharing and coordination that enable swift investigations and prosecution of illegal alien voting.” DHS echoed the DOJ’s view, calling such information sharing “essential to protecting America’s election process by keeping noncitizens off voter rolls” and “an easy step to secure our elections.”
Sus said that by pooling voters’ sensitive personal identifiers, along with their party and voting histories, the administration has “run roughshod” over privacy laws and is creating a “1984-style database” that could be used to surveil political participation.
Balliet, the former HSI agent, expressed similar concerns, saying that handing over voter rolls to run through DHS’ systems amounted to “a mass data-collection effort by the government against its own citizens in a non-criminal setting.” On a practical level, he added, the data sharing increased the risk of breaches that can open law-abiding citizens up to identity theft.
“As a citizen, I want the government to take seriously the protection of my private data,” he said. “If it falls into the wrong hands, it’s not the government that suffers. It’s me.”
Collapsing Claims
By January 2026, scores of HSI agents were fanning out across the country to investigate thousands of leads about illegal voting.
About 25,000 of them involved people SAVE had identified as potential noncitizens on state voter rolls. Another 15,000 were given to HSI’s cross-border financial crimes unit, which was told to prioritize them over money laundering and fraud cases, a former high-level HSI official said. Quiñones’ unit led a separate effort to find additional cases by reviewing whether people who’d recently become citizens had voted before they were naturalized, then lied about it in naturalization interviews.
Still more leads came from running public voter rolls through the HSI lab’s databases, though it’s not clear how many. Agency sources told ProPublica that the rolls of New Jersey and Pennsylvania were searched using the Athena Toolbox. When DHS announced it had found more than 250,000 noncitizens on the registered voter lists of four Democrat-led states, it said that included “as many as” 35,152 in New Jersey and 14,576 in Pennsylvania. Both states have asked DHS to provide the source of the numbers.
Once federal investigators started digging into these cases, however, they often fell apart.
Through May, prosecutors had charged fewer than four dozen people with crimes related to noncitizen voting, ProPublica’s analysis of DOJ data and federal court filings shows, getting convictions or guilty pleas in 14 cases.
The largest cluster of cases was brought in the Southern District of Florida, headed by Jason Reding Quiñones (no relation to the HSI agent), a staunch Trump ally who has led investigations of the president’s political opponents. Voters have been sentenced in eight of the office’s cases; seven others are pending.
Prosecutors have initiated a total of seven noncitizen voting cases as of May in the four states where DHS claimed to have found more than a quarter million noncitizens on the rolls. Collectively, these states have more than 40 million registered voters.
Overall, the paltry numbers reflect the realities HSI investigators confronted on the ground as they tried to convert leads into cases, a ProPublica review of records obtained by voting-advocacy groups Campaign Legal Center and Democracy Forward shows.
Emails between agents and local election officials in Texas and Ohio indicate much of the federal information on purported noncitizens was inaccurate, the consequence of flawed data matching. When agents demanded voting histories and registration forms, they often discovered people weren’t registered in the counties where HSI thought they were. When voters register, they must attest that they are U.S. citizens; agents found some, however, who’d been put on voter rolls in error after disclosing they weren’t citizens. Many had never voted.
Most HSI agents, steeped in pursuing complex international crimes, had no experience in pursuing voting fraud. In his email to Nick Davis at the Justice Department, Teirab, the top deputy in Minnesota’s U.S. attorney’s office, described the chaos that ensued when the state and at least six counties got subpoenas for voters’ records. The demands “presented a host of issues,” Teirab wrote to Davis. (Davis did not respond to a request for comment.)
“An unknown number of agents are constantly arriving in Minnesota without any knowledge of the investigation, the issues already discussed and decided, and the processes they need to follow,” Teirab complained. “Instead of getting up to speed and learning, they have demanded many calls and subpoenas, most of which demonstrate a complete lack of understanding of the investigation.”
In Stearns County in central Minnesota, HSI agents subpoenaed voting histories on 13 people, county officials told ProPublica. Only six turned out to live there.
Minnesota Secretary of State Steve Simon, a Democrat, said the effort revealed federal investigators’ ignorance about “the nuance of elections administration.” The rate of purposeful election crime is “microscopic,” he said. “The very human errors that can happen are not some sort of plot or scheme to dirty up the voting rolls, but are just everyday human beings making mistakes, not intending at all to violate any law.”
As of May, only one illegal voting case has been referred for prosecution in Minnesota. The state has more than 3.8 million registered voters.
Even Republican-led states like Ohio, which has shared its unredacted voter rolls with the DOJ and run them through the SAVE system, have struggled to meet Trump administration demands to deliver cases.
In October, the Ohio secretary of state referred more than 1,000 voters identified as possible noncitizens to the federal government, claiming 167 had voted at least once since 2018. At least nine HSI agents in Cincinnati and Cleveland began to investigate, emails obtained by Campaign Legal Center and shared with ProPublica show.
The agents bombarded local election officials with requests for voting histories and other records, some dating back decades. “Sorry again for piecemealing these to you,” one agent wrote to an official in Butler County, outside Cincinnati, after making five separate requests over two weeks. Mohamed Al-Hamdani, a Democratic member of Montgomery County’s Board of Elections, which received similar inquiries, said they amounted to “witch hunts” that ate up staff time.
In February, Davis sent an email pressing Ohio’s two U.S. attorneys’ offices for an “election integrity” update, asking: “How close are we to complaints/indictments, how many subjects, what issues you’ve run into, how many referrals have been closed and why, how helpful HSI has been.”
As of May, just two voting cases had been referred for prosecution in Ohio and no one had been charged, federal data shows. The state has nearly 8 million voters.
Now, however, U.S. attorneys’ offices are taking a different approach, pursuing prosecutions that could result in prison terms or deportation.
In May 2025, federal prosecutors in Florida filed illegal voting charges against a mother and daughter, both Ukrainian citizens, who’d registered after becoming permanent U.S. residents, then voted in the 2024 presidential election. They’ve pleaded not guilty, with the mother saying in an interview with investigators that she thought having a green card meant she could vote. They and their attorneys didn’t respond to requests for comment from ProPublica. The prosecutor handling the case for the Southern District of Florida also didn’t respond to questions. The mother and daughter are scheduled to go to trial in September.
“Far-Reaching Consequences”
Experts on both elections and national security warn there may be profound longer-term costs to unleashing the federal government’s investigative and prosecutorial might to try to prove the president’s claims about noncitizen voting.
Current and former HSI agents say critical investigations — including task forces aimed at drug rings, human trafficking and money laundering — have languished as the agency has pivoted to take on noncitizen voting and immigration enforcement.
In February 2026, according to court files and emails reviewed by ProPublica, an HSI agent in Ohio who’d been leading a multistate child sex abuse investigation was abruptly assigned to pursue dozens of leads on suspected noncitizen voters, work that took months. It wasn’t until May that he made an arrest in the other case. Prosecutors have filed child exploitation and pornography charges against the man, 38, in connection with acts involving at least five 14-year-old girls.
Cases like this “don’t just stop,” said Balliet, the 23-year HSI veteran. “When you pull people off them for an extended period of time, those criminal networks will adapt, and people get killed and kidnapped.”
Federal data hints at a broader slowdown in what had been HSI’s most urgent work. According to an annual government report, the number of wiretaps obtained by DHS — often to investigate drug trafficking — plummeted to 23 in 2025, from 133 the previous year. In the report, DHS attributed the decline to several factors, including “changes in administration enforcement and prosecutorial priorities.”
“It’s never been a better time to be involved in transnational organized crime,” said John Tobon, a high-ranking HSI official who retired in early 2025.
In its statement, DHS denied HSI has neglected its core mission, saying the agency has increased its efforts to combat criminal cartels, gangs and drug traffickers in response to multiple Trump executive orders.
Though HSI’s voting-related investigations have yielded only a small number of prosecutions, voter advocacy groups worry the Trump administration will use them to justify more extreme interventions.
“Those concerns are playing out in real time, with far-reaching consequences,” said Dax Goldstein, a director at the States United Democracy Center, a nonprofit group that works to build confidence in elections. “When federal agencies are misused and weaponized, people lose trust in government itself.”
Sure, the town’s insurance will pay most of this, but it’s probably fair to say most of this small county’s population (11,823 at last count) wasn’t supportive of local law enforcement’s decision to violate as much of the First Amendment and Fourth Amendment as possible in the apparent hopes of making one local business owner happy.
The backstory is long and convoluted. It involves a local business owner who wanted a liquor license (but had been cited/arrested from drunk driving), her estranged spouse (who shared some documents with Marion County Record reporters), a local attorney who just wasn’t up to the job, a police chief who was far too friendly with the business owner, and a bunch of other law enforcement agencies that pitched in with the constitutional violations just because. And all of that is on top of then-police chief Cody’s animus towards the paper, due to its reporting on his past misconduct.
There’s Kari Newell, a local business person who was seeking a liquor license for a new business when her previous drunk-driving record became public. There’s County Attorney Joel Ensey, who claimed to have no knowledge of the raid until public records showed he actually knew plenty about it beforehand. There’s the Kansas Bureau of Investigation, which also disavowed all knowledge, until it became clear it had knowledge as well, at which point it began publicly condemning Chief Cody and his department. There’s the mayor who didn’t like his deputy mayor and seemed to be all too willing to indulge the police chief. There’s the judge who signed off on the search warrants without reading them and then tried to distance herself from actions — a judge who apparently had some drunk driving problems of her own. There are the communications Chief Cody made to Kari Newell, informing her he was going to raid the newspaper to shut down its coverage of her and, presumably, any further investigation into his law enforcement past. In the middle of all of this, there’s some bullshit computer crime charges, which were invoked despite the newspaper accessing driver record data legally through a third party.
Following the raids — that’s right, raids — multiple lawsuits were filed. Not only did the Marion County PD (under the leadership of Gideon Cody, whose past misconduct was being examined by the local paper) talk a local court into blessing its raid of the newspaper’s office, it also raided the home of the paper’s owner, 98-year-old Joan Meyer, who died less than 48 hours after the raid.
Pretty much every lawsuit filed by the victims of these raids has paid out. Last November, the county agreed to pay $3 million to three of the affected journalists and one of the town’s city council members, who was also subjected to an illegal raid by local officers.
That large settlement followed a $235,000 settlement the town agreed to pay to Marion County reporter Deb Gruver, whose computer was seized along with her personal cell phone by local law enforcement officers.
That brings the total to $3,235,000 (at minimum!). The latest settlement — one that will be paid to another of the paper’s reporters — now means county residents will be asked to contribute to a tab that has now surpassed the $4 million mark:
Phyllis Zorn, the reporter whose acts of journalism served as an excuse for the August 2023 police raid of the Marion County Record, will get $850,000 from the city of Marion to settle her federal lawsuit over the raid.
This is the sort of thing that should make all good Americans yank on their imaginary suspenders and get their rural lawyer shtick on. We should — as a unified drawl — make it clear that we, as the collective “small town lawyer,” think this is some disturbing bullshit. And we should fervently argue in favor of large settlements, even if we know this just means our fellow Americans will be asked to give a little more the next time the budget’s on the agenda.
While I do think its sucks that the public has to pay the price for government malfeasance, things like this encourage more people to vote with their wallets. That doesn’t mean funneling money into some PAC. That means protecting your earnings by expelling the people who were on board with these blatant violations of constitutional rights. And Marion County residents aren’t done paying for the perverse acts of prosecutors, law enforcement, and the mayor who had the cops’ back all the way through this debacle:
The county government is making things right. That it’s doing this with other people’s money doesn’t mean the settlements are meaningless. We, the people, should simply rejoice in our magnanimous nature. Even though we’re getting fucked, we can at least take heart in the fact that we were invited to the climax.
Even as the wider EU was grinding through a long, drawn-out process to figure out which regulatory levers to pull on kids’ safety online, France decided to YOLO it earlier this summer by jumping at the chance to ban all social media for kids under 15. This kind of thing has become popular with out of touch adults in the grips of an ongoing moral panic, since Australia led the way after gambling companies pushed the ban as an alternative to banning gambling ads. Australia’s ban isn’t doing very well, with the majority of kids figuring out how to work around it, and those being left out being the most marginalized and in need of community.
And yet, countries around the globe have all suddenly decided — some based on reading Jonathan Haidt’s badly reasoned book — that they must do this too.
France was the first in the EU, with President Emmanuel Macron gleefully treating France’s willingness to rush in with little thought or understanding as a selling point:
“France is leading the way in Europe in protecting our children and our teenagers,” Macron said. “We will keep on going.”
He wanted the ban to be implemented in mere months, just as kids returned to school.
But that’s all run into a bit of a stumbling block known as the French Constitutional Council, which has said the ban is an unconstitutional attack on kids’ rights to free expression and communication. The Council also flagged a second problem: you can’t enforce an under-15 ban without making every user, adults included, prove their age. The law demanded exactly that, without defining a single condition, limit, or technical standard for how that verification would work.
A court in France on Friday shot downa bill seeking to ban access to social media for under-15s from September — a major blow to President Emmanuel Macron that raises fundamental questions about efforts to protect kids on the internet.
The Constitutional Council, which reviews the constitutionality of French legislation, said the restrictions in the bill disproportionately infringe on minors’ right to freedom of expression and communication.
Reading through the Google translation of the actual ruling, it’s pretty short and to the point. Similar to how the First Amendment requires any restriction on speech to be narrowly tailored to a specific government interest, here the Council says the ban is way too broad and not based on any specific, narrowly defined harm, though it admits that the aims of protecting children are certainly legitimate:
Furthermore, while the established prohibition does not apply to online encyclopedias, educational or scientific directories, or platforms for developing and sharing free software or open-source educational digital projects, the exceptions provided for in the contested provisions remain limited. In particular, these exemptions do not cover collaborative services for sharing leisure, information, or mutual aid content, online communication applications, or online games with strong collaborative and social features, nor do they cover online social networks which, while not inherently educational, are created in connection with educational activities.
Thus, the prohibition established is likely to apply to online communication services whose risks to the health and safety of minors, relating in particular to their content or mode of operation, are not established.
In short, the law goes way too far in issuing a blanket ban of all children, no matter what the circumstances. If you have a legitimate, well-defined problem, come up with a narrowly tailored solution. The French government rushed this one through with little concern for things like that.
It also leaves little room for parents to decide for their own kids what’s appropriate:
… neither the contested provisions nor any other provision sets out the conditions under which the holders of parental authority or the legal representative of the minor, duly informed of the potential risks and safeguards presented by the services concerned, may, in the child’s best interests and in the exercise of their duties under the law, decide to lift the prohibition, limit its scope or authorise access to certain services.
Thus, the prohibition established does not give rise to any particular assessment of the risk to the minor, taking into account in particular his age, his degree of maturity, his family situation as well as the nature of the service concerned.
Also, the age verification attack on privacy is important to recognize:
By prohibiting access for any minor under fifteen years of age to certain online services, the contested provisions imply, in themselves, that any person, even an adult, must prove their age before accessing them.
The Council further notes that the law makes no real effort to figure out how one might implement age verification in a manner that protects the privacy rights of adults.
Of course, having hitched his own legacy to this thing, there’s no way Macron lets it go quietly:
In a statement late Friday, the French presidential office said the government would not be giving up on the bill. It has set a new target date for spring 2027, which coincides with when Macron will leave office.
The statement said Macron “has instructed the Prime Minister to work, as quickly as possible, on a legally sound draft that takes into account” the court’s decision. The ruling hinted at what would make the age restriction align with fundamental rights: giving parents more flexibility.
Politico also spoke to Peter Craddock, a Brussels-based attorney who works on social media regulation, who notes that any other EU country attempting a similar blanket ban is likely to run into exactly the same wall:
“The reasoning is actually equally relevant internationally, throughout the EU, because this fundamental freedom is not specific to France,” he said.
Which is a useful reminder that the freedom of expression problem here isn’t a quirk of French constitutional law. It’s baked into the whole approach — and no amount of “but it’s for the children” framing makes it go away. That’s even more true of the problems with age verification requiring the scanning of everyone’s ID, which is an even touchier subject in large parts of the EU than elsewhere.
Really, though, the bigger, more important message here should be to slow down. What’s incredible is that for all of the political and media class whining that social media is some rogue experiment on our children, none of them seem to consider that abruptly trying to block all social media from kids is just as much an experiment, and one that might have equally damaging effects.
Why not wait and see how the Australian ban actually works in practice? The early results are a mess. I get that Macron and other politicians want headlines and a legacy to point at, but it would be nice if they actually followed what the research shows and looked at how the early experiments of these bans have worked out.
So what France produced here was a total rush job that sacrificed the expression rights of every teenager in the country, the ability of parents to make their own judgment calls regarding their own kids’ access to information and — as a cherry on top — the privacy of every adult who would now need to prove their age at the door to the internet. Thankfully, the Constitutional Council caught all three.
memoryOS is a gamified mobile app designed to significantly improve your memory through short, interactive lessons. Guided by two-time World Memory Champion Jonas von Essen, the software teaches proven mnemonic techniques alongside spaced-repetition schedules. It features a digital library of 3D “Virtual Mind Palaces,” which allows users to easily visualize, organize, and store complex information for fast recall. You can get a 1 year expert plan subscription for $50, 3 years for $120, or 5 years for $150.
Note: The Techdirt Deals Store is powered and curated by StackSocial. A portion of all sales from Techdirt Deals helps support Techdirt. The products featured do not reflect endorsements by our editorial team.
