We're watching the death of human persuasion in real time, and it's happening right where you'd least expect it. The op-ed page. For over a century, the opinion section of a newspaper was a sacred, messy sandbox. It was a place where politicians, experts, and regular citizens clashed using their raw wit, personal grief, and hard-earned perspective.
Not anymore. Today, editors are getting flooded with text that feels uncanny. It looks like an opinion. It uses the right vocabulary. But it has no pulse.
If you think newsrooms can easily spot and block this stuff, you're wrong. They're losing the war. Just look at the recent scandals shaking European media. Germany's Tagesspiegel had to yank opinion pieces penned by its own former publisher and editor-in-chief, Stephan-Andreas Casdorff, after discovering undisclosed AI use. Days earlier, Frankfurter Allgemeine Zeitung got busted publishing an op-ed by Mario Voigt, a state premier in Thuringia, that used artificial intelligence to construct its arguments.
When the people running the papers and running for office stop using their own brains to write, the system is broken. You can't just pass a rule to stop AI on opinion pages. The incentive structure of modern media makes it almost impossible to control.
The Fake Expert Epidemic
This isn't a problem about grammar. It's a problem about the relationship between language and thought. When you sit down to write what you believe, the struggle to find the right word forces you to clarify your actual stance. AI removes the struggle. It spits out a pre-chewed, statistically probable version of an opinion.
The result is a flat, uniform sludge. Nordic political news site Altinget recently flagged this trend across its Danish, Norwegian, and Swedish editions. Their opinion editors noted that incoming submissions are increasingly incoherent, impersonal, and lacking concrete details. They issued five strict rules for contributors, telling writers flat out that language and thought are linked, and they want human voices, not machines.
Good luck enforcing that. The tricks people use to bypass editors are getting sophisticated. We aren't just talking about a lazy freelancer copying and pasting from a chatbot. The current crisis involves three distinct types of fraud that are slipping through the cracks.
- The Ghostwritten Politician: High-profile leaders use large language models to draft policy arguments, turning public debate into a battle of automated press releases.
- The Hallucinated Source: Authors use AI to pull historical anecdotes or public data, completely missing the fact that the machine fabricated the quotes.
- The Totally Fabricated Pundit: Outlets unknowingly publish entire columns written by non-existent people with AI-generated headshots. The Mississippi Free Press caught one of these fake authors on their own site.
Even veteran media executives are falling into the trap. Peter vanderMeersch, the former CEO of Mediahuis Ireland, was caught publishing Substack columns that featured completely hallucinated quotes generated by AI. If the people who spent their entire lives running major media conglomerates can't responsibly manage their use of the technology, how can we expect an unpaid guest columnist to do it?
Why Detection Software is a Total Joke
The immediate corporate response to this mess is always the same. Buy software. Hire a tech company to scan submissions.
That strategy fails because AI detectors are fundamentally broken. They operate on probability, looking for patterns in text structure. If a writer uses a generic template, a detector flags human work as robotic. Conversely, a human can tweak a few adjectives in a machine-generated draft and sail right past the software.
Relying on tech tools creates a dangerous false sense of security. It turns editors into bureaucrats who click a button instead of using their editorial intuition.
The real damage happens when publication workflows speed up. An opinion editor at a mid-sized metro paper might review dozens of submissions a day. They don't have forty minutes to cross-verify every single historical reference or trace the cadence of a writer's voice. They look for a compelling hook and a clean layout. AI excels at providing exactly that, masquerading as polish while hiding a total lack of substance.
The Counter-Argument That Makes Things Worse
Some media owners think the solution isn't to ban the tech, but to lean into it publicly. They want to use software to break the "echo chamber" of opinion journalism.
Take the Los Angeles Times, owned by billionaire Patrick Soon-Shiong. The paper introduced an AI tool called "Insight" to analyze its own human-written op-eds. After publishing an article about the visual dangers of AI in documentary filmmaking, the paper's own automated tool stamped a "center-left" label on the piece and generated an automated counter-argument underneath it, using tech from Perplexity and Particle News.
The tool argued that AI democratizes storytelling, running its text right below the human essay without any prior review by LA Times journalists.
The LA Times Guild fought back immediately, pointing out that unvetted automated analysis erodes public trust. They're right. When a media outlet uses an algorithm to grade and counter its own writers, it stops being a curator of human thought. It becomes a content farm arguing with itself. It tells the reader that human perspective is just a data point to be balanced out by an automated counter-weight.
How Newsrooms Can Actually Fight Back
You can't police the writer's keyboard. You can't sit in their office and watch them type. If someone wants to use a machine to write their op-ed, they will find a way to hide it.
Publishers need to change what they value. If your opinion page prioritizes clean, predictable prose that checks a bunch of partisan boxes, you will get buried in machine text. To survive, editors must pivot to styles that software cannot replicate.
First, stop accepting opinion pieces from people who refuse to do original reporting or share lived experiences. If an op-ed is just a collection of thoughts on a national news story, a bot can write it better. Demand specific, localized details. Ask for names, personal choices, and unique interactions.
Second, pick up the phone. If a piece looks too perfect, call the writer. Ask them to explain their argument in two sentences over a live call. You can instantly tell the difference between a person who wrestled with a complex topic and someone who entered a prompt into a text box.
Finally, normalize transparency over absolute bans. Outlets like Altinget are on the right track by encouraging open disclosure. If someone used a tool to sort their initial data, fine. Let the reader see that. But the actual reasoning must belong to the human whose name is on the byline.
Go look at your favorite opinion section right now. Read the guest columns closely. If the writing feels incredibly smooth, perfectly balanced, and slightly hollow, you're probably reading a machine. Stop giving those pieces your attention. Demand writers who are willing to be messy, opinionated, and distinctly human.
