Mind and Iron: The Four AI Stories That Are Defining This Year

Hi and welcome back to another fine episode of Mind and Iron. I’m Steven Zeitchik, veteran of The Washington Post and Los Angeles Times, senior editor of tech and politics at The Hollywood Reporter and lead turntablist of this newsy album-release party.

Every Thursday we come at you with all the AI and future-world news you need to be informed. Join our journey.

I’d neglected to mention last week, but I will this week, that I recently had the opportunity to chat on-air with the lovely Rachael Myrow from KQED, taking listener calls on AI, creativity and the future. Give a listen if you get a moment.

And speaking of creativity, this week for THR I also had the pleasure of talking to the leaders of OpenArt, a newish startup looking to bring AI tools to the creative world, including a directing tool that basically guides you from ideas to postproduction all through your laptop. They were delightful and interesting, and love or feel conflicted about what they’re doing, their energy is undeniable. You can check out our interview here.

An info-rich issue this week’s as we conduct our semi-annual report card/crystal ball on how AI and our techno-future year are shaping up. We’re coming to the halfway mark of 2026 in a few days, and there have been some storylines, man. From Anthropic’s dance with the Defense Department, to the AI bubble-burst that may finally be looming, to drug discovery that is legit mindblowing, it’s been a heady six months, and the back-end promises to be just as dramatic. We’ll break down what you need to know.

Also, last week we told you about the stakes of the Alex Bores Congressional election in New York. This week we got a result. More on all that and where the AI regulation movement goes from here as well.

First, the quote of the week:

—Alex Bores, sounding hopeful

Quick note o’ summer schedulin’ that we’ll be out next week for the Fourth (/Taylor Swift wedding), back the following week, out the week after, but then with you through much of the summer until a little well-earned break at the end of August. Weather’s hot and pool’s a’beckoning; why don’t you spend some time splashing around with us?

Also, this issue marks the completion of the third year of this here Substack. Woohoo! Thanks for being such an instrumental part of the journey; you guys are what make it all worthwhile.

Let’s get to the messy business of building the future.

China, bubbles and drugs; Bores bows

1. SIX MONTHS INTO WHAT HAS ALREADY BEEN A DIZZYING YEAR WE WANTED TO take stock of what’s been happening on the AI front. So here are four storylines that have already been full of some wild turns, with no doubt plenty more to come.

AI and the Defense Department

How much should the U.S. military be using AI? How much should AI companies allow themselves to be used? The issue came to a head this winter when Anthropic wanted guardrails on what the Defense Department could use its products for, particularly regarding mass surveillance of U.S. citizens and autonomous weapons (said department wanted zero strings),

Anthropic pulled out of the deal, to no small amount of retributive swagger on the Defense Department’s part; the courts stopped that swagger dead in its tracks; but none of it really mattered anyway because OpenAI stepped in to offer its services. Their tools are not regarded as strongly effective but they still get the job done, which means the military got what it wanted and now has a wide range of tools to spy and sift through data as it sees fit.

Just a few weeks ago, the U.S. government ordered Anthropic to cut off access to its powerful Fable and Mythos models for all foreign nationals, a decision that left the company not the least bit pleased. The beat goes on.

Nor is this the endpoint of the military and AI. The Defense Department already relies heavily on Gemini, the Chief Digital and Artificial Intelligence Office has “unleashed” an “AI-first War Department” in which “Warfighters have the AI-enabled capabilities necessary to train, fight and win” (seriously, check out this almost comically over-the-top Web site), and the president earlier this month gave a 90-day timeline to defense and intelligence departments to “responsibly accelerate the use of AI” and rewrite an autonomous-weapons directive in a way that will surely increase its adoption.

The U.S. military is now going full roadrunner on AI, believing these tools and techniques can’t get adopted fast enough. (A Democratic senator from Arizona is trying to get transparency from the Pentagon on this process.) Will these systems actually make us safer? Or just become fodder for exploitation? The rest of 2026 will be fraught and revealing.

AI and the Bubble

We’ve been hearing about a bubble-burst for nearly a year now. The prediction template goes roughly like this:

Companies are spending nearly $600 billion annually on AI investment, and the stock market has risen 8x on the insatiable appetite for chips and other products associated with this speculative boom. When the speculation doesn’t materialize and demand plummets, the stocks that have benefited will crater, the AI bubble will pop and the economy will come crashing down in its wake. So goes the prediction. The losses are too great, the wealth too concentrated, the economy too lopsided and the hype too out of whack with the consumer utility for anything but a dark day to come.

Or as the tech journalist and author Cory Doctorow said in a speech several months ago, “When the AI investment mania halts, most of those models are going to disappear, because it just won’t be economical to keep the data-centers running...The collapse of the AI bubble is going to be ugly. Seven AI companies currently account for more than a third of the stock market, and they endlessly pass around the same $100b IOU….AI is the asbestos in the walls of our technological society, stuffed there with wild abandon by a finance sector and tech monopolists run amok. We will be excavating it for a generation or more.”

Jeremiah, perhaps. But Jeremiah was also mostly right. This week a sell-off on Tuesday brought the NASDAQ down 2%; overseas didn’t fare much better. Some individual tech stocks dropped double digit-percentages (a bit of rebounding happened since). Doctorow spoke to Arts Technica this week to promote his new nonfiction book “Reverse Centaur” (the title refers to a doom prophecy of humans being an appendage to the central AI body) and he decried the bubble and predicted its popping, while outside the interview walls that narrative inched forward.

“Imaginary markets have no agreed-upon valuation because you just made them up,” Doctorow said. “Unless you can turn an imaginary market into a real market pretty quickly, you need to come up with another imaginary market and announce that this is the new imaginary market you’re going to conquer. It’s easier than you’d think because the capital markets have the object permanence of a toddler, and they would lose a game of peekaboo if they were drafted to play in the league. So you can say, ‘Oh, actually, it’s not metaverse. It’s crypto. It’s not crypto. It’s Web3. It’s not Web3. It’s something else.’ And the markets will forgive you, provided you do it quickly enough.

“But something different happened with AI. It is much, much bigger in terms of capitalization than anything we’ve ever seen—not just bigger than other tech bubbles, bigger than other bubbles. When I wrote the book, capital expenditure (CapEx) globally was $700 billion, now it’s $1.4 trillion. Meta wasted $60 billion on the metaverse. They spent $150 billion in the last three years on AI, and they say they’re going to spend another $150 billion this year.”

Doctorow calls for a puncturing of the bubble (“pop it, as quickly as we can,” he said in the speech.) Which could cause plenty of pain. More desirable is a slow deflation, one that neuters the companies that both drive up costs in a fevered economy and use AI to justify the laying off of large swaths of workers — but doesn’t do it so fast it collapses the stock market and hundreds of millions of retirement accounts with it. A less easy trick, and one I’m not sure we could pull off. More likely Doctorow is right and the bubble keeps inflating until it pops, harshly. The next six months will tell.

AI and Medical Discoveries

Having given you the bleak realms, time to turn to a brighter one. The medical use cases for AI have been abundant for a while, and they just keep coming. This week we got another one: a paper in Nature in which AI (specifically deep learning) found a potential leading suspect in a common cause of death.

Some 350,000 people with few risk factors drop dead of sudden cardiac arrest in the U.S. every year, puzzling scientists. The machine found a correlation to cardiac fibrosis, aka a condition in which scar tissue is scattered throughout the heart. That’s something researchers could literally not have done, examining the echocardiograms of hundreds of thousands of people and cross-referencing them with death certificates, and the fact that machine-intelligence could uncover a culprit hiding in plain sight is straight-up nirvana.

Nor is this an idle finding: these deaths can be prevented with an implantable defibrillator. Before, we’d never put one into an empirically healthy person. Now, with this new risk factor uncovered? We just might, and save thousands of lives.

Such discoveries are hardly anomalous. AI recently combed through a huge library of 45 million anti-bacterial compounds — a painstaking and underfunded operation when researched the human way — to successfully find two potential new antibiotics with relative ease. And a paper in the Journal of the American Chemical Society earlier this month laid out how AI found a pocket in a protein that’s a target of cancer treatment that human researchers has missed, potentially allowing drugs to be more targeted and this more effective/less side effect-y.

The ability of machines to comb through so much data and spit out a result may seem like it’s coming to take a low-level job, or claiming designs on a higher-level job it can’t do all that well. But it also is coming to do a job no one can — and give us life and health in the process.

AI and the China Invasion

You know how American tech companies/libertarian politicians use competition from China as an excuse for a complete lack of regulatory oversight? Well, there is a complete lack of regulatory oversight — and China is surging anyway.

A head-spinning number of developments have unfolded just over the past few weeks, and they all favor the Middle Kingdom in the race to dominate AI.

First Microsoft said it was debating building Deepseek — that cheaper AI that back in the winter of 2025 sent shock waves through Silicon Valley — into Copilot, a potential major inroad for China into the U.S. market. Then came another Chinese startup model, Z.ai, which was swelling in popularity and giving the big American companies a run for their money. In fact, the latest rankings for most-used LLMs have Chinese-based products occupying all four of the top spots, unprecedented since the rankings started. Deepseek is so feeling the burn that it just announced it’s doubling staff.

The reason this is happening is not plain technological superiority — and it’s certainly not American regulation, as again, it’s non-existent. No, the main reasons China has suddenly gained an advantage is that they have figured out how to produce AI outputs cheaper, which means it’s cheaper for customers to buy. The country also (not unrelatedly) has kept nearly all its models open-source, which means others can iterate upon it, generally making products cheaper (and better). Silicon Valley firms won’t do that because it cuts into the money they make; if others can build off it they need you less.

Essentially what the Chinese companies have done is trade added profits for the few in exchange for power for the nation. That is a much harder road for an American capitalistic society to mandate, though companies could of course decide on their own to go open-source and counter this rise. (They won’t.) So here we are in a looming age of Chinese AI dominance.

What does this mean for you? Nothing. And everything. The nothing comes because it doesn’t really matter to individual customers where any tech product comes from as long as it’s appealing, and if you doubt that ask people how much they’ve given up TikTok to go all in on Reels.

But Chinese AI superiority also means an American weakening, never good for the economy or general national prospects. Plus the ongoing concerns of what Chinese AI companies could be feeding back into their national-security maw; all this is happening amid worries, for instance, that Anthropic has given access to its powerful/feared Mythos model to a South Korean company, SK Telecom, with ties to China.

On the other hand, if the bubble bursts, at least there will be fewer American giants collapsing under it. Silver linings.

2. FIRST, THE NEWS PART: ALEX BORES LOST.

The AI regulation hawk we told you about last week didn’t lose by a lot — about 4,000 votes out of ~100,000 votes tallied. But still, only good enough for second place behind veteran pol/operative Micah Lasher.

Now, the analysis part. Did Bores lose because of his harsh stance on AI, did his candidacy rise as a result of his harsh stance on AI, or did the issue have no effect at all?

The answer is yes.

First off, there’s no doubt he caught a wave. A 35-year-old Assembly member fairly new to politics running against a 44-year-old Assembly member who’s been in politics for 20 years (and had the backing of some of the biggest names in the state), well you don’t go toe to toe without some real heat on a popular issue. And Bores had that. People are worried about AI companies running unchecked, and Bores’ rise reflects that to a T.

Or as he put it in his concession speech, “Americans everywhere are seeing AI work its way into every aspect of their lives and their economy and they’re looking to see who will stand up for them.”

That said, many, many millions of dollars came in against him, from a super PAC underwritten by OpenAI, Palantir and a16z principals, and the commercials they blanketed the airwaves with, shameless as they were, seemed to have an effect. If the hopeful inference from Bores’ near-win is that a lot of voters vibed with his message, the not-so-hopeful inference from his narrow loss is that tech companies’ resistance to being regulated is even stronger than we thought, and they’re going to back up the Brinks truck to stop even one person who holds some basic common-sense ideas about putting up guardrails.

And finally, the no-effect factor. As much as AI regulation should be a major issue in political campaigns, it has yet to rise near the top of the list of Most Important Issues for voters, which means a candidate cannot ride it and it alone to victory, even in a pretty plugged-in place like Manhattan.

But that doesn’t mean it won’t change. As Bores noted, “Future victories will be built on the progress of this campaign; that’s how movements work.” Bores’ surge as a result of his AI platform has modeled an approach for many other candidates — for people in November contests and people in future contests and frankly even people already in office. Bores has taken an issue that a year ago most Americans would’ve shrugged off as a niche matter beyond their ken and turned it into something urgently relevant to them, to us. at the ballot box and beyond. In that sense, Bores’ candidacy is an indisputable triumph.

Every week we bring you the TSAS — the TOTALLY SCIENTIFIC APOCALYPSE SCORE (tm). It’s a barometer of the biggest future-world news of the week, from a sink-to-our-doom -5 or -6 to a life-is-great +5 or +6 the other way. 2025 ended on an up note, but the score for the year was dismal — a horrendous -42.5. Can we turn things around in 2026? Last week, we occupied positive territory for the third consecutive week. This week? More good news.

AI-MILITARY ENGAGEMENT, AI CHINA DOMINANCE, AI BUBBLES BURSTING — BUT AT LEAST WE’LL LIVE LONGER: -2.5

ALEX BORES SETS A TEMPLATE: +3.5