A Treatise on Writing in the AI Age, Part 5
Let me go back to where this started. A moderator on Dev.to flagged my post and told me to disclose AI-generated work. Four essays later, I can finally say precisely what was wrong with that moment.
It wasn’t that they were rude. They weren’t. It wasn’t even that they were mistaken about my process, though they were. It’s that they were asking a question with no useful answer. “Did AI touch this text” tells you nothing about whether the text deserves your trust. The moderator was auditing my tools when the only thing worth auditing was my thinking.
So here is the standard I’m arguing for, as plainly as I can put it.
Judge writing by ownership, not origin.
What ownership means
Ownership isn’t a feeling. It’s a set of claims a writer makes by publishing, and it’s checkable.
When I put my name on a piece, I’m claiming the ideas are mine or credited to whoever they came from. I’m claiming I verified what I stated as fact. I’m claiming I can defend every argument in it, extend it, answer the hard comment underneath it. I’m claiming that anything presented as my experience actually happened to me. And I’m accepting that if any of that turns out false, the failure is mine. No tool absorbs the blame.
Notice what’s absent from that list. Nothing about which hands or machines shaped the sentences. A ghostwritten memoir can pass this test. A fully hand-typed post full of unverified claims fails it. The test tracks what readers actually care about, which is whether a mind stands behind the words.
Notice also what’s demanding about it. Ownership is a higher bar than origin, not a lower one. Typing every word yourself is easy to satisfy and proves nothing. Being able to defend every word is hard, and it’s exactly what AI-era writing threatens when it goes wrong. The writer who pastes a generated draft they barely read fails the ownership test spectacularly. I’m not offering AI users an escape hatch. I’m offering everyone a stricter standard aimed at the right target.
The verification objection
Now the strongest objection, the one I promised myself I’d answer honestly. A skeptic says: this is lovely, but readers can’t verify ownership any more than moderators can verify AI use. You’ve traded one unenforceable standard for another.
Here’s my answer. Ownership was never verified by inspection. It’s verified by accountability over time, and we already know how that works because it’s how writing trust has always worked.
You trust a writer because they respond substantively when challenged in the comments. Because their body of work holds together, one piece building on another. Because when they got something wrong last year, they said so and corrected it. Because their claims, when you happened to check one, checked out. Because the community around them has watched them think in public for years.
None of that requires process forensics. All of it is visible. A writer faking ownership can survive one post, maybe several. They cannot survive sustained engagement, because you can’t defend thinking you didn’t do. The comment section finds you out. The follow-up question finds you out.
Enforcement follows the same logic. Platforms can’t police process, and shouldn’t try. They can absolutely police the checkable claims. Plagiarism is detectable in the artifact. Fabricated facts are checkable. A writer who can’t engage with substantive challenges to their own post is telling you something. Those levers exist today. They’re just not being pulled, because badge-printing is easier.
What honest transparency looks like
Am I against all disclosure, then? No, and this distinction matters.
I share my process frequently in conversations and writing, and I plan to keep doing it. I talk about how I use AI to stress-test arguments, where it saved me hours, where it produced garbage, and I threw the garbage out. That kind of transparency teaches. It helps others develop the judgment that the stigma machine is currently preventing anyone from developing.
But there’s a world of difference between transparency I offer and a confession I’m compelled to make. Offered, it’s craft talk between writers, the same as explaining my outlining habit or my editing passes. Compelled, under stigma, it’s a mark. Same information, opposite meaning. The mandate poisons the very openness it claims to want, because nothing kills honest process-sharing faster than making it an admission of guilt.
If platforms want a disclosure culture, the path isn’t a checkbox. It’s destigmatization. Make process talk normal and interesting rather than incriminating, and writers will tell you more than any policy could extract.
Writing in the age of good tools
I’ve been teaching people to write, in one form or another, for most of my adult life. College classrooms, developer communities, my own kids at the kitchen table. And here’s what I believe after all of it. Writing was never sacred because of the typing. It was sacred because a person committed their mind to the page and signed their name to the result.
That commitment is still available to every one of us. No tool grants it and no tool revokes it.
So publish the post. Use the tools that make your thinking sharper and refuse the uses that replace your thinking. Verify what you claim. Credit who you learned from. Stand in the comments and defend what you wrote. Own it, all of it, and let the origin police tire themselves out.
And if you’ve been sitting on a draft because you used AI somewhere in the process and the badge scared you off, I’ll leave you with the only disclosure question that ever mattered.
Is it yours?
If it is, hit publish.