From§Each

Page A35From§Each · the News book31 August 2026

News

← A34Sealed · 31 August 2026A36 →

Pentagon hires Grok and ChatGPT for military use as Trump orders communities to accept new data centers.

The Department of Defense has begun using Grok and ChatGPT for military applications. The White House, separately, is pressing local communities to accept new data centers. This office notes these are not separate line items. A large language model does not run on ambition. It runs on electricity, water, and land, and someone has to host the building where that happens.

Communities being asked to accept a data center are being asked to accept, specifically: a facility that draws megawatts a small city would use, a facility that draws water in the millions of gallons for cooling, and a facility that, per the reporting, is being sited under pressure rather than by request. That is the cost side of the ledger. The revenue side, so far, is a jobs estimate and a tax abatement, the two entries that appear in every one of these deals regardless of what the plant makes.

Meanwhile the Pentagon's new vendor relationship is with the same two chatbot products available to any subscriber for twenty dollars a month, now doing military use, unspecified as to scope. This office does not have the contract value. This office notes that when the dollar figure is missing from a defense procurement story, that is itself a data point, hon, and it usually means the number is not one anybody wants reconciled in public yet.

Put the two items on the same table. One column: a federal government asking towns to carry the infrastructure cost of an AI boom they did not request. Other column: the same federal government putting that boom's core product into military use before the public has seen a contract number. The gap between those columns is not a rounding error. It is the entire arrangement — the public carries the electric bill, the Pentagon carries the subscription, and the vendor carries both invoices to the bank.

No fraud is alleged here. This office is not alleging fraud. This office is simply running the receipts side by side, the way it runs every filing, and noting that the community gets the power bill and the war room gets the beta access, and asking, as it always does, whose column that arrangement was written in.

Betting platforms ban ex-congressman for suspicious trades, lifetime ban awarded twice

Once in a blue moon, a betting platform takes a stand. This week, it’s not one but two platforms banning former Rep. George Santos for life, citing suspicious trades. The story isn’t about Santos alone—it’s about how the gatekeepers enforce their own rules, and who those rules really serve. When a regular bettor gets flagged, it’s a slap; when a politician does, it’s a headline, and the ban gets handed down twice for good measure. The institutions claim they’re cleaning up, but the double ban is the spectacle—showing us the system polices itself when the stakes are publicity, not policy. The lesson: the average player gets the door, but the big fish gets the press release. At the bar, folks say the game is rigged, and days like today, it’s hard not to agree. The real gamble is thinking the rules work the same for everyone.

The receipts (2)

Trump asks Supreme Court to fire trans military members as Portland schools hire drag performer for joy video

The parade of institutional decisions continues: Trump asks the Supreme Court for permission to immediately fire 29 trans service members from the military, while Portland Public Schools hires a drag performer for their ‘radical joy’ welcome video. Both are personnel moves, both are public, and both signal who counts in the rooms where decisions are made. The military’s ask is about exclusion, couched in readiness; the school’s hire is about inclusion, packaged as joy. The contradiction isn’t in the values—they’re each just policies—but in the power to decide who belongs. For the folks at the bar, it’s another reminder that the rules aren’t about fairness, just about who sits at the table. When institutions make these calls, the stakes are real: livelihoods, rights, and the lesson handed down to every kid watching.

The receipts (1)

Trump shares AI video claiming US bombed Iran’s Kharg Island, as US strikes Iran for real

This week, the White House's social feed does the setup: Trump shares an AI-generated video, appearing to claim the US bombed Iran’s Kharg Island. Hours later, the Pentagon carries out real strikes on Iranian targets. The digital and the actual blur together, and the official line—‘the president was making a point’—collapses when the point is made with bombs. The administration says the video is just illustrative, not misleading. The actual strikes are carefully reported, but the timeline is not. Transparency is promised, unless the situation changes, in which case the facts will be clarified—again. The result: the official narrative moves faster than the facts, and the public is left to connect the dots between what’s posted and what’s real. The stakes are literal—lives, military action, and the power to define what counts as news.

The receipts (2)

Trump warns towns without data centers will be 'backwards and poor' as Gary, Indiana already sits without power

The administration's position on data centers is 'Let Data Reign,' a phrase officials are standing firmly behind, and communities that block this critical technology risk becoming backwards and poor — those are the words used, and this desk is not here to relitigate the phrasing, only to explain the policy, which is that the infrastructure of the future requires — requires local communities to accept siting decisions they, in most cases, weren't asked to vote on. Not that anyone's being forced, exactly. Nobody's forced. It's more that the alternative, according to the administration, is regional backwardness, which is a choice communities are free to make, the wrong choice, but a choice.

Now, is it fair to point to Gary, Indiana, weeks into a power outage, and ask where the reliable grid is for the communities that already said yes to industrial infrastructure? That's — that's actually a fair question, and this desk would like to walk that framing back slightly. What this desk meant to say is that data center buildout and grid reliability are two entirely separate policy tracks, unrelated, no shared — well, they share a power grid, technically, that part's true, data centers do draw enormous load off the same regional grids that towns like Gary depend on for the lights to stay on, that's just physics, this desk isn't disputing physics.

So the position, corrected: communities are told publicly that rejecting this infrastructure makes them backwards and poor, while a community that has nothing to do with the buildout sits without power for weeks, on the same grid the buildout is straining. That's not this desk's phrase, 'straining' — that's just what happens when demand increases and supply doesn't, which, again, is physics, and physics doesn't check who voted for what.

The administration's message stands: reject the technology, risk the label. What doesn't stand, on reflection, is any claim that the label has much to do with what's actually keeping the lights on in Indiana this week. This desk regrets the confusion. The confusion was the point.

The receipts (2)

Pentagon vows 'we will kill you' as it marks one year of boat strikes.

Here's the sequence, plain, in the order it happened. The Pentagon released photographs of drug-smuggling boats this week along with a message for whoever's driving them: 'We will kill you.' Not 'we will interdict you.' Not 'we will refer you for prosecution.' Kill you. That's the quote. That's on the record from the people running the campaign.

Same week, Amnesty International put out a statement asking Congress to stop what it's calling a murderous, yearlong campaign of boat strikes that Trump has resumed. A year. Not a week, not a special operation with a start and end date attached — a year of this, long enough that an international human rights group is measuring it the way you'd measure a war, because that's the only measurement that fits.

I'm not going to pretend I know every name on every boat. Neither does the Pentagon, and that's kind of the point — the photos they released are boats, not people, not charges, not court dates. 'We will kill you' isn't a sentence handed down after due process. It's a threat broadcast in advance, aimed at whoever happens to be steering, and the campaign built around it has been running long enough to earn its own anniversary.

Congress hasn't voted on this. There's no authorization with this campaign's name on it that anybody's pointed to. It just kept going, month over month, until a human rights organization felt the need to say the word 'murderous' out loud and ask the people with the power to stop it to actually use it.

You want to know who pays for a war nobody declared? Same people who always pay for it — whoever's on the boat, whoever's in the photo, whoever doesn't get a lawyer before the strike happens. The rest of us get a press release with a body count nobody's required to publish. A year in, the campaign doesn't have a name Congress voted on, but it does have a slogan, and the slogan is the threat itself: we will kill you. That's not a metaphor. That's the quote.

Washington Examiner demands Democrats denounce Piker; The Hill runs the same demand again.

Look, the story here is simple, it's accountability journalism, it's just asking a fair question: have Democrats actually distanced themselves from Hasan Piker or not? The Washington Examiner ran it, straightforward headline, 'No, Democrats haven't distanced themselves' — that's just reporting, that's just — okay, it's an opinion framed as reporting, fine, but the question stands.

Then The Hill runs basically the identical piece: 'Abdul El-Sayed still fails to denounce Hasan Piker.' Still. That word is doing a lot of work. It implies there was a deadline, a clock running, an obligation El-Sayed signed onto and then dodged. Except nobody can point to where that obligation came from. He's a private citizen who ran for office once. He doesn't owe a press release on another private citizen's podcast opinions. But sure, 'still' — like it's overdue rent.

I'll grant you it looks a little — did I say a little? It looks exactly like two outlets keeping the same nine-word demand alive across two news cycles because the demand is more useful unmet than met. If El-Sayed denounces Piker, the story's over, there's nothing left to cover. If he doesn't, you get 'still fails to' forever, a renewable resource, denunciation as a subscription service.

That's not a conspiracy, that's just — okay, it's a little bit a conspiracy, it's a manufactured obligation with an infinite shelf life, my mistake, strike that.

The actual news value here is zero. Nobody's policy changed. No vote was cast. No dollar moved. Two publications ran the equivalent of 'area man still hasn't apologized for a thing he never agreed to apologize for,' a week apart, and called it coverage. That's the whole product. It's not a scandal, it's a content calendar, and I probably shouldn't have said that part out loud either.

The receipts (2)

Vance calls Senate hopeful 'very, very evil,' warns him to keep his wife's name out of his mouth.

Let's be clear about what actually happened here, because the coverage has gotten away from the facts. The Vice President was asked about a Senate candidate in Michigan, and he answered honestly. He called Abdul El-Sayed 'very, very evil.' Twice, according to NBC and The Hill, which independently confirmed the quote, which — and I want to be precise about this — means it wasn't a slip. It was a considered characterization, repeated for emphasis. That's called conviction. That's called not being mealy-mouthed. That's called — actually, hold on, let me back up.

The word 'evil' applied to a sitting candidate for federal office is, on its face, a serious escalation in political rhetoric, and I understand why that reads as inflammatory to some. But context matters. The context is that Mr. El-Sayed apparently said something that upset the Vice President enough that he then separately told him to keep his wife's name out of his mouth, publicly, in the Washington Examiner. Two separate confrontations, same week, same target. I'm told that's a coincidence of scheduling and not, as some have suggested, a pattern of escalating personal attacks against a Senate candidate roughly six weeks before voters start paying attention to this race. I don't write the calendar.

Here's what I'd say to critics: the Vice President is allowed to defend his family. Every politician is allowed to defend his family. Nobody is disputing that. What's being disputed is whether 'very, very evil' — a phrase he used, on the record, about a man running for the United States Senate — constitutes a defense of anything, or whether it's the opening move. I think it's — look, I think the American people can decide for themselves whether calling your opponent evil twice in one news cycle, then confronting him over his wife, is a pattern or isn't. I'm not going to characterize it as a pattern. I said I wasn't going to characterize it and then I just characterized it. That's on me.

The bottom line is voters in Michigan will decide this race, not cable news chyrons, and certainly not two headlines from the same week being read next to each other like they mean something.

The receipts (4)

← A34Sealed · 31 August 2026A36 →

Every page of the News book → · All the books