Archive
Every issue of The Fourth Estate AI Brief, in reverse chronological order.
Friday Deep-Dive
Turn a 2-hour meeting into a story draft in 20 minutes
You didn't get into this business to spend Tuesday night at the council meeting and Wednesday morning typing it up. So this week I went looking for the simplest ways AI can hand you back real hours, not toys, actual time. I found five, and today we go deep on the biggest one: turning a long public meeting into a solid story draft, with the fact-checking built right in so your name is never on a robot's mistake. The exact steps, the exact prompt, and the four other time-savers are below.
Friday, August 7, 2026
Tuesday Brief
The $1.5B Settlement That Isn't Really a Win
The interesting thing this week with AI and publishing — the headline numbers don't tell the story. Anthropic is paying $1.5 billion to authors. That sounds like accountability. But dig into the math and you find authors getting $3,000 per pirated book while the company's valued at $965 billion. The fine is 0.2% of Anthropic's worth. Meanwhile in New Mexico, an AI news site published a fake flood death because nobody was checking. And in Australia, publishers are pushing for AI licensing fees before the same thing happens there. The pattern: AI companies are moving fast, the guardrails are lagging, and community publishers are caught in the middle.
Tuesday, August 4, 2026
Friday Deep-Dive
Claude can now learn a task by watching you do it once
A Stanford AI founder posted a 30-second video last week that stopped me cold. He showed Claude how to pull names out of a messy PDF, talking through it out loud the way you'd train a new hire, and Claude turned that one demonstration into a reusable tool it can run forever. No prompt writing. You just do the task once while it watches. One thing to know before you get excited: it's Mac only right now. I went looking for the button on my own Windows machine this week and it isn't there. So read this one as a heads up, not a homework assignment.
Friday, July 31, 2026
Tuesday Brief
Two Courts, Two Verdicts: The AI Copyright Split Widens
The interesting thing this week with AI and publishing — we got a split decision. A U.S. judge approved Anthropic's $1.5 billion copyright settlement while an Indian court ruled in OpenAI's favor against a news agency. Same week, opposite outcomes. For community publishers watching from the sidelines, this is the clearest signal yet that where you're located — and who your lawyers are — matters more than who's right.
Tuesday, July 28, 2026
Friday Deep-Dive
Your hands-free sales assistant just went live
Every AI tool so far has asked you to stop, sit down, and type — which makes them useless in the exact moment a salesperson needs help: in the car, in a lobby, crossing a parking lot with your hands full. This week Anthropic changed that. Claude's rebuilt voice mode can now actually do things while you talk to it — read your email, check your calendar, catch you up on Slack — out loud and hands-free. Below: what changed, five plays for a rep who lives in the field, exactly how to switch it on, and what plan you need (you can start free).
Friday, July 24, 2026
Tuesday Brief
The Paywall Penalty and the Bot Majority
The interesting thing this week with AI and publishing — the rules about what makes content discoverable are being rewritten, and the old playbook isn't just outdated, it's actively working against you. Two studies dropped this week that should change how you think about your website. One shows that bots now outnumber humans in web traffic. The other shows that if you're behind a hard paywall, AI assistants pretend you don't exist. Both point to the same uncomfortable truth: visibility in 2026 isn't about getting clicks anymore. It's about whether the machines even know you're there.
Tuesday, July 21, 2026
Friday Deep-Dive
Why ChatGPT keeps giving you the same answers
An attendee asked me this at a webinar I was presenting on Wednesday. but I've heard the identical frustration from many different people: “Why do I keep getting the same thing back from ChatGPT?” It's the number-one complaint about these tools, and the honest answer is that “the same thing back” is really two different problems with two different fixes. One takes two minutes, once. The other is a small change to how you ask, each time. Get both right — whatever kind of work you do — and the machine stops sounding like a bored stranger. Here's how, in plain steps, for ChatGPT and Claude.
Friday, July 17, 2026
Tuesday Brief
The Washington Post's AI Defense, Claude's Free Window, and the Personalization Trap
The interesting thing this week with AI and publishing — the defensive plays are starting. For months we've watched AI platforms quietly siphon readers away from publisher sites. Now the countermoves are arriving. The Washington Post's tech arm launched a tool to keep reader questions on your site instead of ChatGPT. A journalism student proved AI personalization works but hallucinates quotes. And Anthropic bought publishers another week to test Claude's best model for free. The theme: AI isn't coming for publishing anymore. It's here. The question is whether you're playing offense or defense.
Tuesday, July 14, 2026
Friday Deep-Dive
GPT-5.5 wrote your ads. GPT-5.6 thinks like a creative director.
OpenAI released GPT-5.6 yesterday, after a federal review held it up — the second time this summer Washington has sat in the middle of a frontier model launch. You'll see plenty of coverage about benchmarks and pricing. Here's the part that actually matters if you sell advertising: the new model stops and thinks before it writes. Ask the old one for an ad and it wrote you an ad. Ask this one properly, and it interviews you about the advertiser, challenges your assumptions, and hands back three complete campaign concepts with layout notes and designer instructions. That's not a copywriter. That's a creative director — and this week I'm going to walk you through turning it into yours.
Friday, July 10, 2026
Tuesday Brief
The AI Licensing Gap Is Already Here
The interesting thing this week with AI and publishing — the deals are getting signed, and we're not at the table. A Brookings report dropped this week showing that AI licensing revenue is flowing almost exclusively to the biggest publishers. Microsoft cut its first Australian media deal with Nine Media. Meanwhile, click-through rates from AI systems collapsed six-fold by Q4 2025. The math is brutal: large publishers get paid, local publishers get scraped, and the traffic we used to get from search is evaporating. The window to organize is now, before these deal structures become permanent.
Tuesday, July 7, 2026
Friday Deep-Dive
The AI That Clocks In Before You Do
At a small paper, everybody wears six hats — and nobody has a spare hour. So here's the shift worth your Friday: both Claude and ChatGPT can now run a task on a schedule, on their own, while you're doing something else. You write the instructions once — check these three sites every weekday at 6 a.m. and email me what's new — and the AI clocks in without you, does the job, and leaves the result in your inbox. This week's deep dive is a practical one: how to set a scheduled task up in each tool, a dozen ideas sorted by who does what at your paper — publisher, editor, writer, ad rep, designer — and the honest version of what to expect before you trust it with your Monday morning.
Friday, July 3, 2026
Tuesday Brief
The Local Press Goes to Court, Google Plays Hardball, and Asia Builds Its Own AI
The interesting thing this week with AI and publishing — the local press finally stopped waiting for someone else to fight the copyright battle. Nearly 400 community newspapers filed suit against OpenAI and Microsoft on Tuesday. Meanwhile, Google is telling publishers to hand over training rights or lose their Showcase payments. And overseas, Asian AI labs are shipping frontier models while Anthropic's export ban drags on. Three stories, one theme: the rules of who owns what — and who gets to use what — are being written right now.
Tuesday, June 30, 2026
Friday Deep-Dive
Claude doesn't just write anymore — it builds. Here's a tool for your readers.
For two years, AI has handed you words — drafts, summaries, answers. This week, the bigger shift came into focus: AI that hands you a working thing instead. Anthropic made headlines launching "Artifacts" for its engineering tool — live, shareable pages that build and update themselves. That part is for software teams. But here's what got lost in the noise: a simpler version of the very same idea is already sitting inside the regular Claude you use, and a reporter with zero technical skill can use it today to build something genuinely useful for readers. So this week, instead of "watch this space," we're going to actually build something — a simple calculator your readers will love — and I'll walk you through it step by step.
Friday, June 26, 2026
Tuesday Brief
Your Journalism Is Already Training the Machines
The interesting thing this week with AI and publishing — your content is already the biggest source feeding AI answers, and you probably didn't sign up for that. A Muck Rack study found journalism accounts for 27% of all AI citations, beating out corporate blogs, Wikipedia, and everything else. Meanwhile, readers are starting to ask ChatGPT about your town instead of visiting your site. The data this week paints a picture of an industry whose work product is being harvested at scale while the harvesters build tools that could redirect traffic away from the source. Here's what showed up that matters for publishers like us.
Tuesday, June 23, 2026
Friday Deep-Dive
Soon, you'll own your AI. Here's how to taste it this weekend.
Two weeks ago the U.S. government pulled a top AI off the market; last week a Chinese lab answered by giving away one nearly as good, free, to the whole world. But here's what most people miss: it isn't just one model. There's a whole shelf of free, "open" AI models now — from China, yes, but also from Google, Meta, and Europe — and any of them you can download and run on your own computer, with no subscription and no data ever leaving the building. We're not all the way there yet, and I'll be straight with you about the price of the giant flagship. But my honest bet is that by the end of this year, a model good enough for real newsroom work will run on hardware a small paper can actually afford. This is the heads-up — plus a click-through with the exact, step-by-step setup so you can try it this weekend.
Friday, June 19, 2026
Tuesday Brief
New York Legislature Passed AI Disclosure Law Now Awaiting Governor Signature — Here's What That Means for the Rest of Us
The interesting thing this week with AI and publishing — New York legislature passed the first state law requiring news outlets to label AI-generated content. Not a guideline, not a best practice, a legal mandate. Governor Hochul hasn't signed it yet, but the bipartisan vote signals where this is heading. This week's brief also covers a German court ruling that makes Google liable for AI lies and ChatGPT's upcoming update.
Tuesday, June 16, 2026
Friday Deep-Dive
Well. That aged in one day.
So this is a first — and I owe you a quick correction, because the newsletter I sent you Friday morning is already out of date!!!
Monday, June 15, 2026
Friday Deep-Dive
Anthropic's next-tier AI just landed in your account — free until June 22
On Monday, Anthropic released Claude Fable 5 — and this one is genuinely different. It's the first model from a tier ABOVE the Opus class ever released to the public, built on the same technology Anthropic itself spent months warning was getting too powerful to ship. They shipped it anyway, wrapped in new safeguards — and right now it's included free on every paid Claude plan, but only until June 22. Here's what the leap actually is, who gets it, the honest fine print nobody else is covering, and what a publisher should do with it in the next ten days.
Friday, June 12, 2026
Tuesday Brief
The Internet's New Customer Is an AI Agent — Plus McClatchy's Ultimatum and a Quiet SCOTUS Win for AI
This week one story towers over the rest, so we're leading with it. There's a shift happening on the internet that almost nobody in local news is talking about: for thirty years the user of the web was a human being. That's ending — AI agents are becoming the customer, and what that means for your website, your classifieds, and your subscription page is the most important thing in this issue. Below it: a legacy chain telling its newsrooms to embrace AI or watch the company struggle, a quiet Supreme Court ruling that handed AI companies a liability shield, and 900 designers on what they're actually doing day-to-day.
Tuesday, June 9, 2026
Friday Deep-Dive
How to get the exact image in your head out of ChatGPT
Everyone who's used ChatGPT for image generation has had this experience: you ask for something, you picture exactly what you want, the AI gives you something different — and you accept it, because you don't know how to tell it what was actually in your head. Most of the AI images you see online are the result of that compromise. They're not the image the creator wanted; they're the image ChatGPT defaulted to. The skill that separates "AI is fine, I guess" from "AI gave me exactly what I pictured" is something you can learn in twenty minutes — and here it is.
Friday, June 5, 2026
Tuesday Brief
The AI Language Problem Nobody's Talking About
The interesting thing this week with AI and publishing — we're starting to see the homogenization problem show up in actual research. A study out this week confirms what some of us have suspected: AI-assisted newswriting is making journalism sound the same everywhere. Same phrases. Same structures. Same voice. For community papers, where local voice is the whole product, this matters more than the latest chatbot update.
Tuesday, June 2, 2026
Friday Deep-Dive
Stop pasting your work into Claude
Until recently, using Claude meant a lot of copy-paste. The article you wanted summarized, the email thread you wanted analyzed, the spreadsheet you wanted explained — every conversation started with manually pulling content out of the source and pasting it into the chat. Anthropic just changed that with Connectors: a feature buried in your Claude settings that, once enabled, lets Claude read your Gmail, your Drive, your Notion workspace, your Calendar, and a few dozen other tools directly. It's the moment Claude stops being a faster typewriter and starts being an actual research assistant with access to your archive.
Friday, May 29, 2026
Tuesday Brief
Google Blows Up Search, Publishers Block the Archive, and Sports Teams Show Us What Not to Do
The interesting thing this week with AI and publishing — we're watching the ground shift under three different parts of our business all at once. Google announced the biggest change to search in 25 years. More than 340 local papers are blocking the Internet Archive. And professional sports teams are fumbling their AI rollouts in ways that should make every newsroom editor pay attention. None of these stories exist in isolation. They're all connected by the same question we keep coming back to: Who controls what happens to our content, and who benefits when AI systems consume it? Let's work through what actually matters here.
Tuesday, May 26, 2026
Friday Deep-Dive
The AI command that doesn't quit
OpenAI shipped a command inside Codex called /goal that changes what AI can actually do for you. You give it a finish line — measurable, specific — and it loops on itself, trying, testing, fixing, retrying, for hours or days until it gets there. It's working in Codex desktop right now — though the autocomplete menu doesn't show it, which almost made me miss it entirely. Here's what /goal is, how to access it, and the gotcha you need to know about before you try.
Friday, May 22, 2026
Tuesday Brief
AI Gets a Business Operating System, Google Wants You to Optimize for AI, and Journalism's Burnout Crisis Gets Real Help
The interesting thing this week with AI and publishing — we're watching the tools mature from "things you chat with" into "things that actually do the work." Anthropic shipped 31 pre-built business skills inside Claude this week. Google published an official guide telling publishers how to make their content show up in AI answers. And there's a new mental health helpline specifically for journalists, because apparently 80% of us are burning out (which... yeah, that tracks). Let's dig in.
Tuesday, May 19, 2026
Friday Deep-Dive
Six AI image models. One typed sentence. Zero browser tabs.
I just generated six versions of the same image — from six different AI models — in thirty seconds. One typed sentence into Claude Code, no browser tabs. For anyone in your shop building images on deadline — reporters, ad designers, page compositors — that's the difference between "this will take an hour" and "done before deadline." Below: the 10-minute setup, every screenshot, and four ways to put this to work this week.
Friday, May 15, 2026
Tuesday Brief
Publishers vs. Meta, Copyright Reality Checks, and Digg's AI Comeback
The interesting thing this week with AI and publishing — the legal ground is shifting faster than most of us can track. Five major publishers just sued Meta for training AI on their books without permission. Meanwhile, courts are issuing contradictory rulings on whether that kind of training is even illegal. And somewhere in the middle of all this, you and I are trying to figure out what we can actually *claim* as ours when AI touches our work. Honestly? This week felt like watching three different games being played on the same field. Let's sort through it.
Tuesday, May 12, 2026
Friday Deep-Dive
Your voice, or theirs?
One piece this week, in the Deep Dive. AI is already in your office whether you've planned for it or not — and the words coming out of your shop are starting to reflect that. The two-hour exercise below is the cheapest insurance you can buy against your subscribers quietly noticing.
Friday, May 8, 2026
Tuesday Brief
The 75% Question: What Google's AI Code Numbers Mean for Your Newsroom
The interesting thing this week with AI and publishing — Google announced that 75% of its code is now AI-generated, up from 25% just eighteen months ago. That's not a typo. This trajectory tells us something important about where the human-plus-AI workflow is heading... and how fast it's getting there.
Tuesday, May 5, 2026
Friday Deep-Dive
A quick note, since this is the first one.
Welcome to the first edition of "The Fourth Estate" newsletter. Here's what you'll get from me: Tuesdays — the AI news that actually matters for community publishers, in plain language. No hype, no "10 tools that will change everything." Just what shipped, what it means for a paper like yours, and whether it's worth your Tuesday. Fridays — a field-notes piece like this one. Something I tried at my desk this week — the prompt, the workflow, what worked, what didn't, what I'd do differently. The stuff I wish another publisher had told me before I figured it out the hard way. I'm writing this from the desk, not the sidelines. If something here is useful, hit reply and tell me. If something's wrong, hit reply harder.
Friday, May 1, 2026
Tuesday Brief
The Byline Battle: When AI Puts Your Name on Someone Else's Work
McClatchy's decision to slap reporter bylines on AI-repackaged stories sparked union grievances this week—and it's a cautionary tale every small publisher should study before rolling out their own AI workflows. Meanwhile, new research confirms what we suspected: your readers want transparency about AI more than they want innovation.
Tuesday, April 28, 2026