The Centre Daily Times in State College, PA, has voted to unionize after months of pushback against its parent company’s AI tool—a move that, according to The NewsGuild-CWA, makes it the first newsroom in the union to cite AI adoption concerns as a primary reason for organizing.

As Nieman Lab reported, the Centre Daily Times staff voted to join The NewsGuild of Greater Philadelphia last month. All eligible editorial staff signed authorization cards, and McClatchy voluntarily recognized the union. The catalyst, reporters told Nieman Lab, was McClatchy’s Content Scaling Agent (CSA) tool—an AI system that repackages existing articles into short-form summaries for publication or video scripts—and a March internal meeting where Kathy Vetter, McClatchy’s chief of staff for local news, told staff the company would use their bylines on AI-generated content unless union contracts prohibited it.

Josh Moyer, a senior reporter at the Centre Daily Times, took that as a signal. “It was essentially like, if you’re not in a union, your byline gets used; if you are in a union, we’ll follow what the union says,” Moyer told Nieman Lab. “If we want to control what happens to our byline, that’s the company telling us that we need to form a union.”

McClatchy introduced the CSA tool at the paper earlier this year. Reporters initially published at least one CSA-assisted story per week under a generic byline noting AI assistance. But in late February, management changed the policy: AI-generated content would now carry the reporter’s actual name. Reporters objected that it misrepresented their work to readers.

“When our names go on a thing, it says that this article or video is from that person, but that is just not true in this case,” said Trebor Maitin, a service reporter. Maitin was the first reporter at the paper to have his byline changed to reflect AI assistance.

The NewsGuild-CWA’s president, Jon Schleuss, said unionized newsrooms have had more success keeping AI content clearly labeled: “Unionized newsrooms are the ones where McClatchy’s AI slop gets a clear label. In non-union newsrooms, the AI slop may be carrying a real human reporter’s byline.”

Multiple McClatchy publications have seen byline strikes over the CSA tool, and some have taken labor actions over the tool and related workplace issues. For the Centre Daily Times, the union opens the door to formal collective bargaining and the ability to join coordinated actions at sister publications.

“Some of us use AI a lot more and are okay with it,” Maitin said. “But there is an overall understanding that we need to be able to have a say in this, and that unionizing at least gives us a seat at the table.”

The post A newspaper unionized because McClatchy put reporters’ names on AI content appeared first on The Media Copilot.

I’ve been working with Claude Cowork extensively over the past month and a half. And not coincidentally, I’ve found myself accomplishing more during this period than at almost any other time in my career. The shift toward agentic work represents a transformation so fundamental that its impact is difficult to grasp until you actually experience it.

Just one example: As someone running a business that sells AI training courses online, email marketing is an important component of reaching potential customers. But the work itself is tedious: segmenting my email list, creating templates, writing largely similar drafts, and scheduling them in my email provider—a piece of software I look forward to using about as much as a visit to the dentist.

Now I hardly ever touch that software; Claude Cowork does it for me. When you have access to agents, you can loop them in on any computer task with three beautiful words: “You do it.” AI doesn’t just draft emails for me—it puts them in the campaign builder, targets the right audience, gets all the settings right, and then taps me on the shoulder (via a notification) so I can approve the work before it schedules everything to go out. Once you start working with agents, you quickly start crossing things off your to-do list faster than ever before.

Becoming the CEO of your job

This represents more than accelerated productivity. It’s a fundamentally different way of working. Instead of personally grinding through individual tasks, the focus shifts to defining desired outcomes, delegating execution to digital workers, and evaluating their output. Instead of simply doing your job, you become the CEO of it, delegating many tasks to agents.

So what happens to a newsroom when everyone starts working agentically? Over the past 30 years, reporters and editors have needed to become skilled at many different systems: project-management software for tracking stories, content management systems for publishing them, SEO plug-ins, social media management platforms—the list goes on. Agents open the possibility that journalists could instruct them to manage all of this infrastructure while they go and do the important, human-centered work of reporting and editing.

But the complications emerge when this same agent model gets applied to journalism’s core function: writing itself. This came to a head recently with the uproar over what The Plain Dealer, Cleveland’s primary newspaper, is doing: leveraging an AI writing agent so reporters can simply feed notes and context to create stories. To be clear, all the stories are then edited, and the reporter has final say over the copy. But applying agents this way brings up hard questions about jobs, skill-building, and career paths.

Yet beyond this specific scenario lies a broader reality: agents will almost certainly assume much of the repetitive work surrounding content creation and distribution. Whether it’s social media management, SEO (and GEO), or getting all the little drop-down menus, boxes, and tag fields in your CMS just right—those are all jobs for agents. More importantly, roles that are centered around optimizing those tasks will gradually go away.

Consider what happens: when search and social platforms drive audience discovery, newsrooms organize work around those algorithmic preferences. Many roles emerged that were simply writing to a trend, publishing undifferentiated “quick hits” around trending topics to maximize clicks. Those jobs were effectively hyper-optimizing production of formulaic stories, writing for algorithms and chasing virality through pattern recognition. An AI system can accomplish this faster, at higher volume, and more efficiently than any human.

Here’s the paradox: this development might actually prove beneficial for journalism—something I predicted in a column I wrote almost exactly a year ago. Agents are a crucible for knowledge work, burning away anything and everything that can be automated, leaving only the parts of the job that can’t be easily repeated—the work that requires either creating new information or judgment, context, and taste.

The agentic newsroom

If you were designing a newsroom optimized for AI from the start, using this principle, the bulk of positions would focus exclusively on the distinctly human elements: cultivating trust with sources through direct access and personal relationships, conducting original reporting and uncovering information exclusive to your brand, determining which stories resonate most with audiences and which narrative angles matter most, and applying the craft of storytelling across all of it.

While that sounds appealing in certain respects, the economic reality is harder: with agents executing most of the work, fewer jobs will likely exist. In almost all cases, organizations will be smaller, with different career paths, even if the work is richer.

A current limitation is the scope of what agents can access. Tools like Claude Cowork and Claude Code become truly powerful only when they can move beyond drafting and into systems (email, CMS, analytics, internal documents). That is where most organizations get uneasy. Granting an agent permission to act inside those environments raises questions about security and accountability. Most teams are still feeling their way through this, limiting agents to narrow tasks or read-only access. But that tension is temporary. As guardrails improve and familiarity grows, those permissions will expand, and with them, the scope of what agents can do.

The fundamental premise remains unchanged: journalism’s purpose is not threatened. Instead, its true essence becomes visible when machines handle the repetitive parts. An AI-first newsroom doesn’t mean a less human one. In fact, it means the opposite. When the repeatable work is handled by machines, what remains is the work that defines the craft: earning trust, finding new information, and making sense of it for an audience. The uncomfortable part is that there may be fewer people doing that work. The hopeful part is that the work itself becomes more meaningful.

A version of this column appeared in Fast Company.

The post What an agentic newsroom will look like appeared first on The Media Copilot.

A single unreleased AI model has triggered emergency regulatory mobilization on both sides of the Atlantic. UK financial regulators are holding urgent talks with the government’s cybersecurity agency and major banks to assess risks posed by Anthropic’s Claude Mythos Preview — days after US Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened an emergency meeting with Wall Street’s top CEOs over the same concerns.

In the UK, officials from the Bank of England, Financial Conduct Authority, and Treasury are in talks with the National Cyber Security Centre. Representatives from major British banks, insurers, and exchanges are expected to be briefed on cybersecurity risks at a meeting with regulators within the next two weeks, according to Reuters. The BoE, FCA, and NCSC all declined to comment.

The US response was more public. White House national economic adviser Kevin Hassett confirmed on Fox News that Bessent and Powell had convened bank chiefs — including the CEOs of Citigroup, Morgan Stanley, Bank of America, Wells Fargo, and Goldman Sachs — to warn of cyber risks from the model. JPMorgan CEO Jamie Dimon was unable to attend. The urgency of the meeting reflected the capabilities Mythos Preview has demonstrated in controlled testing: the ability to identify and exploit weaknesses across every major operating system and every major web browser.

Anthropic has stopped short of a broad release, citing concerns the model could expose previously unknown cybersecurity vulnerabilities at scale. The company has been navigating an increasingly complex relationship with the broader tech and media ecosystem as its models grow more capable.

What Mythos Preview is — and who can use it

Despite not being publicly available, Claude Mythos Preview is already in active use — under strict controls. Under a program Anthropic calls Project Glasswing, select organizations have been granted access to the model for defensive cybersecurity work. Partners include Amazon, Microsoft, Apple, Google, Nvidia, CrowdStrike, and Palo Alto Networks. Access has since been extended to approximately 40 additional organizations responsible for critical software infrastructure.

Anthropic says Mythos Preview has already found “thousands” of major vulnerabilities in operating systems, web browsers, and other software. The company has committed up to $100 million in usage credits and $4 million in donations to open-source security groups as part of the program.

The framing is defensive. But the same capability that finds vulnerabilities can, by definition, be turned toward exploiting them — which is precisely what regulators appear to be stress-testing.

Why regulators are moving fast

The simultaneous and independent responses from UK and US financial regulators signal that Mythos Preview represents a qualitatively different kind of AI risk than those regulators have previously had to assess. Prior AI regulatory concerns have centered on bias, misinformation, and systemic market risks — as seen in ongoing debates around AI copyright policy and AI use certification. A model with demonstrated offensive capability against critical software infrastructure — in active use, even in a restricted form — is a different category of problem.

It is also a compressed timeline problem. The model exists. It is being used. The regulatory frameworks to manage it are still being assembled.

All three UK agencies — the BoE, FCA, and NCSC — declined to comment on the talks. Anthropic had not responded to a request for comment at the time of the Reuters report.

The post UK and US financial regulators hold emergency meetings over Anthropic’s Claude Mythos appeared first on The Media Copilot.

The Associated Press, founded in the mid-1800s to help New York newspapers share reporting costs, is offering buyouts to an unspecified number of U.S.-based journalists — the latest move in a long-running transformation from wire service to technology data company.

The News Media Guild, which represents AP journalists, said more than 120 staff members received buyout offers on Monday. AP executive editor and senior vice president Julie Pace said the goal is to reduce global headcount by less than 5%, though she acknowledged the cut among U.S. staff would likely exceed that figure depending on how many people accept.

“We’re not a newspaper company and we haven’t been for quite some time,” Pace said.

The numbers back her up. Over the past four years, AP’s newspaper revenue has fallen 25%. Big newspaper publishers, once the organization’s financial foundation, now account for just 10% of income. Gannett and McClatchy both dropped AP in 2024. Lee Enterprises — publisher of The Buffalo News, the St. Louis Post-Dispatch, and the Richmond Times-Dispatch — is now seeking an early exit from a contract due to expire at the end of 2026.

Where the growth is coming from

While the newspaper business contracts, AP’s technology revenue has grown 200% over the same four-year period. Kristin Heitmann, senior vice president and chief revenue officer, put it plainly: “If you can think of a large technology company, they are a customer of ours.”

AP was among the first news organizations to move aggressively into AI deals, agreeing in 2023 to lease part of its text archive to OpenAI. It has since launched on Snowflake Marketplace to license data directly to enterprises, stood up AP Intelligence to sell data to financial and advertising sectors, and last year secured a deal with Google to deliver news through the Gemini chatbot — Google’s first content deal with a news publisher.

Elections data is another growth vector. AP saw a 30% increase in election data customers between the 2020 and 2024 cycles, and last month agreed to sell U.S. elections data to Kalshi, the world’s largest predictions market. ABC, CBS, NBC, and CNN all signed on to the AP elections service last year.

What the restructuring looks like

Beyond the headcount reduction, AP is doubling down on video — it has already doubled the number of U.S. video journalists since 2022 — and deploying rapid-response teams that contribute to major stories regardless of geographic base. The organization says it will maintain a presence in all 50 states.

The union is pushing back. In a statement, the News Media Guild said AP “refuses to offer [staff] appropriate training and tools” and is “flirting with artificial intelligence — ignoring the opportunity to differentiate AP news stories as ones that are and always will be created by human journalists.” The union also said AP declined a request last week to bargain over AI use.

AP did not immediately comment on either claim.

Pace framed the restructuring as a strategic choice made from stability, not distress. “The AP is not in trouble,” she said. “We’re making these changes from a position of strength.”

The post AP offers buyouts as AI and tech companies now drive revenue growth appeared first on The Media Copilot.

Journalism students are more skeptical of AI than their professors expect — and a classroom experiment at Northeastern University is surfacing exactly why that matters for how journalism schools teach the technology.

Dan Kennedy, who teaches a graduate ethics seminar at Northeastern, recently devoted a class to hands-on AI use, asking students to run interview transcripts through Claude and evaluate the results. What he didn’t anticipate: students pushed back harder than he did. “I was surprised to learn that they are as skeptical of AI as I am — maybe more so,” Kennedy wrote in Poynter, noting that he himself regularly uses Claude for source research and brainstorming.

The exercise gave two teams the same transcript — an interview with Tracy Baim of the LGBTQ+ Media Mapping Project — and asked them to generate bullet points, a 600-word summary, a news story, a headline, and a social media post. Students then evaluated each output for accuracy, utility, and ethical disclosure requirements. The bullet points came back too long; the news story was serviceable but flat; the headline Claude auto-generated was judged weaker than the one students explicitly requested.

The discussion questions Kennedy designed cut to the core tensions in AI-assisted journalism: Is it accurate? Is it better than what a human would produce? Is it worth the time saved? And what does disclosure actually require?

One question that generated the most friction: a policy at Cleveland.com and The Plain Dealer, where editor Chris Quinn has reporters submit notes to AI, which then drafts the story for human review before publication. Kennedy asked students whether that practice is ethical if disclosed. The answers, he wrote, were “thoughtful, nuanced” — which is another way of saying the students didn’t let him off easy.

The experiment points to something journalism educators are grappling with across the country: the gap between teaching students about AI and teaching them to use it critically. Kennedy’s approach — put the tool in students’ hands, make them evaluate outputs against specific ethical criteria, then discuss — is closer to the latter. It also surfaces a real tension: students entering the field now are skeptical of AI in ways that may conflict with newsroom practices they’ll encounter on day one.

What Kennedy’s class doesn’t yet account for, by his own admission, is the coming cohort of students who grew up with generative AI as a baseline assumption. How they’ll engage with these same questions — and whether their skepticism will look different — remains an open experiment.

The post Journalism students are more skeptical of AI than their professors appeared first on The Media Copilot.

Of the 440 applications Texas Tribune chief product officer Darla Cameron received for an AI engineering role, 90% were junk — many apparently written using the very tools candidates were supposed to understand. One application literally read: “Here’s a short response that’ll work for this.”

That anecdote, from Hacks/Hackers and Poynter’s AIxJournalism Day at SXSW, captures the split-screen reality of AI in newsrooms right now: genuinely useful when deployed against real problems, already producing absurd side effects when cargo-culted. Poynter’s Alex Mahadevan spent the day listening to newsroom builders — not theorists — and distilled five lessons worth reading carefully.

Start with a specific pain point. Every AI tool actually being used in newsrooms started with a complaint, not a mandate. Pew Research Center’s audience team was spending 95% of its time writing formulaic social posts and 5% engaging with audiences. A WordPress plugin now drafts the formulaic posts automatically; the team reviews options and moves on. That freed time goes to reading comments and responding to what readers actually ask. “Where is the mundane bullshit work that you’re sick of doing?” said Upasna Gautam of the News Product Alliance. “That is a great pathway.”

Draw the line between AI for thinking and AI for writing. Nebraska Public Media’s chief innovation officer Chad Davis stopped using AI for writing entirely — not good enough. But he uses it constantly as a “curiosity partner” for research and brainstorming. His labs team uses AI for concept art and music prototyping: instead of pitching an idea in a meeting, they show a working prototype. Most newsrooms haven’t drawn that line clearly. If you don’t decide, your staff will make their own inconsistent rules.

Be specific about where the human sits. KQED tested AI to identify notable clips from their hourlong radio show Forum. “I’m not ready to say the AI can choose the four most notable moments,” said editor-in-chief Ethan Toven-Lindsey. “But if you put a producer in the loop to make sure those are the right moments, that felt doable.” SWR, a German public broadcaster, has community managers review AI-flagged comments before anything gets acted on. “Human in the loop” is too vague — the newsrooms getting results name exactly where.

Use AI to get closer to audiences, not further away. The Texas Tribune trained a chatbot on its school voucher coverage. When readers asked questions the Tribune’s reporting hadn’t addressed, reporters got story ideas. Pew studied what creators do differently from traditional newsrooms and found a feedback loop: they read comments, answer questions, and build new content from what audiences ask. Any AI strategy that doesn’t free up staff time for that kind of direct audience engagement is probably solving the wrong problem.

Learn to build things yourself. This is the one that cuts across every other lesson. Mahadevan — not a software engineer — built a fact-checking research tool prototype for PolitiFact on vacation using AI coding assistants. Agentic coding tools like Claude Code have lowered the floor dramatically. Trei Brundrett of New_Public said it reminds him of the early web, when gatekeeping vanished and anyone could just build. “The newsroom people who pick up that skill are going to have an outsized advantage,” Mahadevan writes. The ones waiting for someone else to build it for them will be waiting a long time.

The post Five lessons from newsrooms that stopped talking about AI and started building with it appeared first on The Media Copilot.

Investigative journalism is hard, expensive, time-consuming, and often dangerous. I’ve been sued, jailed in Cuba for spying, and even kidnapped over my 40-year career.

But AI is turning the tables and putting new powers in my hands. I’d go so far as to say it’s democratising investigative journalism by giving those powers to anyone.

It’s a big claim, so let me break it down in a literal field report and reveal how I’m doing the accountability reporting of an entire newsroom with a laptop and a $30-a-month subscription. AI has been crucial in how I was able to build a 19,000-strong newsletter audience in 99 countries, which led to me being invited to address the UN and advise multiple governments… in less than three years.

Let’s begin by busting a myth. This isn’t getting the AI to write 1,700 versions of the same article and blasting it across the socials. Nope, this is the opposite. Using the AI to do the grunt work, freeing me up to do the rest.

Used the right way, AI shifts the asymmetry in publishers’ favour. For decades, Big Tech has sent armies of lawyers, comms teams, and lobbyists to control the narrative. Journalists have been left to fight with notebooks and deadline pressure. The information gap was a moat, and tech knew it.

But now, it’s draining. A reporter with the right prompts can now process documents at a speed that used to require billion-dollar resources. Journalists can now do the digging they were trained for, and use the tech to turn it into hard-hitting reporting.

Now a single journalist can hold Google or Meta, Iran or Russia, history or political doublespeak to account—and still have time for lunch.

Finding the needle

I have sources like journalists always have, but many of mine are no longer human. I have alerts set up in search, and notifications on court papers and SEC filings. My tipoffs come in as a steady stream 24 hours a day. Many are nothing, but then one is a trigger. It happened the other week.

The judge in charge of breaking Google’s search monopoly ruled that a technical committee must be established to do the job. Everyone wants to know who they are.

An alert popped up on my phone from an obscure automated court reporting AI that the three had been named. The link gave me the court papers. I was off.

I dropped the committee members a line on LinkedIn, used AI to research their careers and found years-old articles that hinted at their personalities. Within an hour, I profiled them and broke the story. Then I sat down to write it. Boom. Job done.

An antitrust lawyer whose name had been linked with the role rang me from New York saying he’d scoured the court papers and couldn’t find the names. They were buried, I told him. Deep. Even with all his resources he couldn’t find it with his team. Now we’re connected too.

Financial forensics

Apples and pears. That’s journalists and accountants. Journos do words and geeks do maths. Only I love both, because you need to follow the money to find the truth.

Every quarter, the tech firms report their earnings to the US Securities and Exchange Commission. They are interminable—dull and data-laden, but full of gold. Best of all, the tech titans who love nothing more than privacy have to put these numbers out in the open to satisfy their obligations to shareholders.

It’s a goddam turkey shoot. I have uploaded years of financial filings, shareholder updates, Wall Street earnings calls into my own small language model. I’ve trained it on my historic reporting, so it knows what I am looking for. For example, is the 90.5%* of advertising Google sends to its own search and YouTube rising or falling? Spoiler: It’s always rising.

That’s why the open web is in danger of collapsing. It’s why publishers have no money to fund newsrooms. It’s also why experienced single operators like me can strike.

Apple’s numbers told me it’s reliant on China amid a tariff war. Meta’s told me 97.6% of its income relies on ads. Snap showed it relies on selling ads to the youngest teens.

These data points that justify headlines are often buried in footnotes, YOY comparisons that used to take weeks and geeks to reveal—but not now.

The fact-checking machine

AI’s a brilliant sub-editor after you’ve taught it your style.

My SLM—I call it RoboRicky—then reads the draft and alerts me if I’ve forgotten a relevant fact in a previous article. It even suggests charts and images. My newsletter contains more than three million words of investigative journalism now, but there’s more than 10 million words of source material in RoboRicky.

It checks every word against the source material to confirm it’s accurate and flags anything that it thinks is wrong.

I’ve also had fun using Google’s Gemini to punch holes in its CEO’s fibs, and had OpenAI run an analysis on whether its deals to buy content were fair. (They weren’t.)

RoboRicky + my brain + my instincts + a superhumanly unwise amount of coffee now power my one-man newsroom. No team. No budget. Me, a laptop, and my killer AI pal.

*Footnote: RoboRicky corrected an error. I’d said Google sent 89% of its ads to search and YouTube. The actual number from the Q4 SEC filing last month (and not reported but calculable by doing the complex maths) was even more, 90.5%.

The post AI is making the one-man newsroom a reality appeared first on The Media Copilot.

If you’ve been even half-watching AI lately, you’ve probably run into Matt Shumer’s “Something Big Is Happening” essay,or, at minimum, the tidal wave of takes it kicked up. Shumer’s basic claim is simple: his own coding workflow has shifted from writing code to prompting, reviewing, and signing off on AI output that’s close enough to “done” to feel uncanny. It’s framed as a warning to knowledge workers everywhere: AI has effectively absorbed my job, and yours is next.

There’s already a small library’s worth of response essays picking apart what Shumer gets right and where he leaps too far, and I’m not trying to add another spine to the shelf. But journalism is knowledge work, too, and it recently had its own—slightly less viral—brush with the same existential questions.

The editor of Cleveland.com (a.k.a. The Cleveland Plain Dealer), Chris Quinn, wrote a column describing how a college student who had applied for a reporting job withdrew their application when they found out how the publication uses AI. Besides leveraging the tech to help generate story ideas, the newsroom developed an “AI rewrite specialist” to write stories based on the material that reporters gather. By ditching writing, according to Quinn, their reporters have been able to reclaim an extra workday each week.

The backlash was predictably vicious. On X, Axios reporter Sam Allard earned a lot of likes by comparing what Cleveland.com is doing to being an “AI content farmer,” while various veteran journalists on Substack expressed various degrees of outrage and dismay. Most of the reaction was along the lines of this piece from journalist Stacey Woelfel: “Writing is an integral part of the reporting process.”

The newsroom’s new fault line

That last line is true, but it’s also not the whole story. What Quinn describes can’t be waved away quite so cleanly, because newsrooms have been unbundling reporting work for decades. Reporters regularly collaborate on one article, with one person taking the lead on the draft while others supply interviews, documents, and context; nobody argues the supporting reporters somehow didn’t do “real” reporting. And in breaking-news moments, reporters often text, email, or phone in their notes to an editor or writer who turns the raw feed into publishable copy.

We all understand, at least implicitly, that reporting and writing aren’t the same skill—even if the best journalists make them feel inseparable. What Quinn and Cleveland.com seem to be doing is using AI to make that separation explicit, formal, and scalable.

This also fits the popular, almost comforting story people tell about “responsible” AI in the workplace: let machines take the repeatable work they can do faster, so humans can spend their limited hours on the parts that actually require judgment and presence. For reporters, that’s the human stuff: calling sources, learning what’s new, asking the second question, and earning trust over time.

And here’s the uncomfortable part: AI is now legitimately good at writing. A lot of what we’ve seen over the past few years hasn’t helped its literary reputation (yes, we’re all tired of the rampant em-dashes and the “it’s not X—it’s Y” bits). But if you use the strongest models—and you’re even mildly intentional about prompting and editing—they can deliver clean, coherent, competent prose.

If we’re being honest, “competent prose” is exactly what a large chunk of daily news requires. Many, if not most, reported stories are built to transmit basic information about what happened, with minimal interpretation, and they’re often written in AP style—a set of constraints that’s effectively a template. It’s not quite code, but it’s functional writing, optimized for speed, clarity, and accuracy. The job is to get the facts right, add context, and move.

Seen that way, the reporter isn’t removed from the process so much as repositioned inside it. Shumer describes becoming a supervisor to an AI building machine; journalists may find themselves supervising writing bots, making sure a story is shaped correctly out of the material they’ve gathered. In Quinn’s newsroom, reporters have final say over the copy.

What gets lost when nobody writes

None of this guarantees a happy ending. Some writers can’t report, some reporters can’t write, and plenty of people are good at both. So what happens when the job is redesigned to force a choice? Do you become a feature or opinion writer, where voice and craft are the value, or do you specialize in the reporting side and let an “AI rewrite specialist” (or whatever comes next) handle the draft?

This leads to the biggest worry: skill-building. Even if Quinn is right and this system truly buys back time, how do junior journalists become better writers if they aren’t writing every day? When Woelfel says writing is integral to reporting, I think he means it’s integral to storytelling—the act of deciding what matters, what comes first, what gets emphasized, and what gets left out, all in service of an audience. That’s curation and prioritization as much as expression.

This is the point Ben Affleck was getting at when he drew his famous line between AI as a craftsman and AI as an artist. Craft can be taught, outsourced, templated; artistry is harder to mechanize. But it’s also hard to become an artist if you never get reps as a craftsperson.

The irony of Shumer’s essay is that even as it argues AI will soon disrupt most knowledge work—and even name-checks journalism as an industry in the crosshairs—it’s written in a distinctly human voice. I honestly don’t know if he used AI to fully or partially write the piece, but I’m certain that if he did, he also was meticulous about every word.

That’s the sliver of optimism here. Even if we push some of the craft of writing onto machines, we may not lose as much as the most alarmed reactions assume. Audiences still want a human touch; if that touch moves upstream—from drafting sentences to shaping the narrative and deciding what’s true and important—it’s still a touch. It’s true that no one wants to read AI slop. But it might turn out that the most valuable reporting skill in the future will be the ability to turn slop into stories.

A version of this column appeared in Fast Company.

The post What critics get wrong about Cleveland.com’s AI rewrite experiment appeared first on The Media Copilot.

Most newsrooms that have adopted AI policies have done something admirable — and insufficient. A new briefing from the Center for News, Technology & Innovation synthesizes 30 peer-reviewed research papers on AI governance and finds that existing policies get the principles right but skip the operational details journalists actually need to follow them.

The CNTI report, released Feb. 17, is the third briefing from the organization’s AI and Journalism Research Working Group. Reviewing policies from 52 global news organizations across 12 countries, researchers found that newsrooms consistently prioritize transparency about AI use, human supervision of AI tools and human verification of outputs. But few policies define what “appropriate” or “proper” AI use actually means in practice.

The gap matters. As one example, AI translation tools can introduce gender biases — assuming doctors are men, nurses are women — that a journalist using a third-party tool may never catch. Existing policies focus on AI outputs, not the systems that produce them, making these subtle errors nearly invisible.

The procurement blind spot is arguably the bigger problem. Researchers found almost no AI policies that address how newsrooms should evaluate, contract with or monitor third-party AI vendors. A 2025 study of 16 AI tool contracts found that most gave developers the right to change terms of service without notice — a risk most individual journalists aren’t even aware of. Meanwhile, newsrooms’ growing reliance on tools built by Google, Microsoft and Amazon deepens their dependence on the same platform companies that already control much of their distribution.

The policy gap isn’t limited to the Global North. A Thomson Reuters Foundation survey of 221 journalists in the Global South found that roughly 80 percent said their newsrooms have no AI policy at all. That number has likely improved since the survey was conducted, but the structural barriers — no access to technical expertise, difficulty getting organizational buy-in, the pace of technological change — haven’t gone away.

The working group’s practical recommendation: treat AI policy development the way you treat coverage decisions. Include people with different job responsibilities and lived experiences. Draw on the lessons of earlier technology policy cycles — photo editing, social media — where the same tension between values and operational specifics played out. And start thinking seriously about procurement: what your AI contracts actually say, who can change them, and whether your organization has the leverage to push back.

For most newsrooms, the answer to that last question is no — but knowing that is the first step toward addressing it.

The post New Research: Newsroom AI policies strong on principles, weak on practice appeared first on The Media Copilot.

Neil Steinberg of the Chicago Sun-Times has made a habit of this. Every February, he asks the latest version of Google’s Gemini to write his column — same prompt, new model — and publishes the results. The 2026 edition is worth reading, not because the AI failed, but because it mostly didn’t.

Gemini 3.0 produced a strong headline (“The Ghost in the Machine is Just Us”) and opened with a scene: Steinberg on the Red Line, watching a young man ask AI to write a poem for his girlfriend. The prose was confident, the tone matched, and the argument landed. The problem? None of it happened. Steinberg never got on the Red Line. There was no young man. The scene was invented — and Gemini delivered it as casually as everything else.

Steinberg asks the actual question plainly: “Will we continue to care if things are true anymore? Must the news have actually happened?”

That’s the question that matters for journalism. The voice problem is largely solved. Gemini 3.0 writes competently in the style of a working print columnist — casual, self-deprecating, city-specific. The fabrication problem is not solved. And fabrication delivered in a convincing voice, at scale, is a categorically different threat than the clumsy AI slop of two years ago.

Steinberg notes, correctly, that AI didn’t create the American appetite for comfortable fiction over inconvenient fact. “We didn’t need AI to undermine the value of truth. Look at who we picked to run the country. Twice.” But AI industrializes the production of plausible-sounding nonsense — and his annual test is a useful benchmark for how fast that industrialization is moving.

For newsrooms, the lesson isn’t that AI can write columns. It’s that the output is getting harder to distinguish from the real thing while the underlying problem — hallucinated specifics presented as lived experience — remains unchanged. The voice got better. The ghost is still lying.

The post A columnist asked Gemini to write his column. The result was good, and that’s the problem appeared first on The Media Copilot.