Medium and the Blanket AI Ban 😶

UPDATED

Thank you, @fatherofcurses for making me stop and actually think through solutions. So I've updated my post on Medium and suggested this system that I think we all would be able to live with (your here is Medium's):

👉 Don’t exclude everybody by default. AI ≠ “bad”. It can be misused, sure, but stereotyping an entire group of writers because of a few bad actors isn’t okay.

Here’s what I propose:

• Create an AI pre-approval system.
• If a writer wants their AI-assisted work to be eligible for your programs, let them apply.
• Get your human reviewers involved — tear the post apart, see if it’s worthy.
• If it passes, give them a badge 🏷️ that automatically shows up on their page.

Now, two good things happen:

1. Writers aren’t forced to keep disclaiming “AI-generated” every post — the badge already says it.
2. Once approved, their work is treated like everyone else’s. No blanket exclusion, just fair review.

Yes, it’s a little more work on your end. But it’s a whole lot easier to swallow — and way more fair — for the rest of us.

---

THE ORIGINAL

I’ve been meaning to write this since I first saw Medium's post on Friday — but, you know, life happens. So a little late, but I wanted to share. You tell me — am I the crazy one here? 🫤

👆 Also, River has shown up in a few posts already — the Leonardo reimagining of my blog character. She’s still missing some key features (training tokens are tapped out until the month resets), but you get the idea. 😆 Expect a couple more tweaks coming soon.

So Friday was my chill day (rare, but they do happen) and I noticed somebody had followed me on Medium. Which is funny, because I’d already canceled my membership — it just hadn’t expired yet. Five bucks a month isn’t exactly breaking the bank, so I thought, fine, maybe I’ll toss a few posts over there alongside LinkedIn.

That was the plan. Until I landed on Medium’s homepage and saw this headline:

“We want your feedback: how can writers use AI to tell human stories?”

🦄 Perfect hook, right? Except two scrolls later I was fuming like I was just called in for a rollback on a Friday night. 😡

---

The Fine Print 🔍

By the time you dig through the post, the documentation, and the linked policies, the message is clear:

• AI as spellcheck? Totally fine. ✅
• ChatGPT outlines and fact-checks? Approved. ✅
• Anything more (drafting, editing, all other AI assistance)? ❌
  • You’re officially outside their definition of “writer.”

And if you do dare use AI, you have to declare it at the very top of your post, within the first two paragraphs, and again on every image. Metadata doesn’t count.

🦄 TL;DR: If you lean on AI for anything beyond grammar, you can write on Medium, but you’ll never get boosted, partnered, or featured. You can post, but you’re standing outside in the rain while everyone else gets a table inside. 🍽️

---

Why It Feels Off 🤔

Look, I get it. Nobody wants a feed full of unedited AI sludge. I’m not trying to game the system either. I’m not chasing claps or badges. I write because it’s fun, because I need a break from code, and because I have opinions the world clearly needs to hear. 😉

AI doesn’t erase my voice — it sharpens it.

So calling everything in between spellcheck and prompt-posting off limits? No appeal, no nuance, just blanket exclusion? That’s where I’m lost.

Developers talk nonstop about AI. Colleges too. It’s a huge, messy, important conversation. And Medium — the site that prides itself on being the “top blogging platform” — takes the most stereotypical, exclusionary stance possible. ChatGPT usually gets painted as the villain here, but really, it’s just the stand-in for all the tools that make the work easier.

Honestly, it feels less like quality control and more like gatekeeping. Nuance would’ve made sense. Even a "don't automate your blog posts" sounds great. This? It's like they skipped gray completely and went straight to Twilight Zone monochrome.

---

My Reply ✍️

You know I couldn’t resist. 🤷‍♀️ Yes — I dropped a response directly on Medium's front door, and linked to it right in their own discussion. That’s the only thing I’ve ever published over there besides the previous post on Blogging with Bots I forwarded as an explanation.

If you want the full-on-rant version of how I pushed back, here it is:

#HumansForAI: A Story of “Little Value” | by Ashley Childress | Aug, 2025 | Medium

Ashley Childress ・ Aug 29, 2025 ・

Medium

---

Wrapping It Up 🌱

I’m calmer now, but the question still bugs me. AI doesn’t erase my voice — it just lets me shape it faster, sharper, and with a little extra spark. So if Medium can’t see that, maybe that’s their loss.

And hey, I don’t want this to be a solo rant echoing into the void either! I want to know what you think. Is this really such a crime against writing, or is Medium being willfully old-school about it?

🦄 Let’s hash it out right here — the corner of the internet where the fun people hang around anyway.

---

🛡️ Powered by espresso shots, a healthy eye-roll, and a touch of AI cleanup ⚔️

All opinions are mine, ChatGPT just helped keep the rant from running off the rails (again).

Comments

[Giorgi Kobaidze]

Aside from the post itself, can we take a moment to appreciate how great your cover images have become? I'm preparing a new article. I know the content will work, but I'm just worried about picking a suitable cover image. I know it’s not a big deal, but the eyes eat first. 😁

[Ashley Childress]

@georgekobaidze — thank you, kind sir. 🙃

This little sub has already paid for itself. Ten or twelve bucks, whatever — worth every penny! My only hang-up is that the style I want lives in Phoenix, and since you can’t train that one yet... yeah, we’ll see if the XLs can fake it. If they get even halfway close, I’ll call it a win 😆

Also, I’ve got a ridiculous pile of notes — mostly so I didn’t have to keep re-not-really-watching the same videos on repeat like some kind of AI-powered Groundhog Day. If you decide to jump in, ping me and I’ll ship a few your way.

And seriously, diffusion prompting ≠ NLP prompting. They are two totally different beasts. I was not at all ready for that when I hit “subscribe.” But hey, this not-at-all-artistic person finally has an outlet for some of the randomness rattling around in my head.

The bad news? All of this is written for art people. Art people know things like lighting and cinematography. The difference between modernism and post-modernism. Me? At best, I’m Googling... but mostly I just copy the big words from the real art people. 🤷‍♀️🤣

[Kelly Emma]

Kelly services jobs Hiring.
Here all details you Complete the applications now. Copy this link and paste any browser. 👇👇👇
is.gd/Xy1vhF

[𒎏Wii 🏳️‍⚧️]

This would have sounded incredibly obvious in 2010, but: If you don't write your own content, you aren't a writer.

If a platform wants to make sure to only serve users content that was actually written by other humans, then maybe that just isn't the right platform for you to publish on.

There's some gray area around the brainstorming process; but drafting? editing? You know what the combination of those two things is? Writing. What medium is banning is using AI to 1. draw two circles 2. draw the rest of the owl.

And that's ignoring the fact that all modern AI is fundamentally based on content theft, and inherently immoral.

[Ashley Childress]

👀 Appreciate the feedback! But hey, if you haven’t read my supporting post How I Blog with Bots (But You Can Still Blame Me), you’re basically arguing about my process with half the instructions missing. 📋🤖

Let’s clear the air (at least enough so we can both breathe):

1. Draft = 100% Me

Every word in that first draft? Mine. Usually dictated over enough lunch breaks to count as a part-time job. The AI doesn’t even get a say until I’m done thinking out loud. If it dares interrupt, it gets a timeout (and my instructions updated accordingly 😆). My thoughts, my voice, my embarrassing tangents from start to finish. Period.

---

2. Editing: Not What You Think

Here’s where everyone gets it wrong: editing with AI isn’t a magical rewrite. It’s me giving ChatGPT a heap of voice notes including:

• half-sentences
• questionable punctuation (if any exists at all)
• a long, complicated, ever-growing list of rules

That’s the part everyone conveniently forgets about with AI: my personalized instructions, spelling out exactly what it’s allowed to touch and—more importantly—what’s strictly off-limits.

The AI is basically my ultra-picky copy editor: it only makes the changes I’ve explicitly told it are allowed. All the personality, all the oddball references, and all the bits that actually sound like me are protected by a growing wall of “do-not-touch” instructions.

Is it perfect? Not even close 😆 Especially since GPT-5 started acting like an intern on their first day and stopped responding to my prompts the same way GPT-4 did. So, yep, I started over. But those instructions get better (and weirder) every round.

So what if the AI screws up (which it does—frequently, and sometimes spectacularly)? Simple: I update the rules and run it again until it sounds like me (well... mostly me on caffeine).

---

Why bother?

Because it’s fun. I enjoy wrangling the chaos and pushing AI to see how far it can go—without losing what makes my writing mine. I get my stories, my voice, and don’t have to care about grammar unless I’m in the mood. Typing? 100% optional. And this reply? Same process, start to finish.

So, tell me which part of that isn't writing? Is it the part where I’m doing all this straight off the top of my head, no notes in sight? Or is it that I’ve standardized my editing, then let AI handle my typos and tangent detours? Or maybe it’s the magic of starting a sentence, abandoning it for a 3-minute rant, and then picking it up right where I left off. My posts would never be this good without that particular style woven in.

---

It's not a secret

The fact that I’m the SME for GitHub Copilot at work? Yeah, that’s a bonus. I do most of this in my spare time. At least half (probably more) of my posts are 100% dedicated to AI because I genuinely enjoy it. Sure, my first few blogs didn’t have the “RAI” footers, but the second I made that connection, I went back and updated every single one—full disclosure, always.

Lately, I’ve been learning about SEO on top of the usual “don’t let me use any word worse than damn” and “highlight every mention of Home Depot so I can personally review it before I hit share.” And just like I said earlier:

AI isn't me being lazy because it's almost always more work than not — that’s me crafting. And I'll be damned if I will let you or anyone else tell me otherwise. 🦾💪

---

One Last Call to Action 🫵

If you’re still skeptical, DM me—I’ll pull the chat conversation for one of my posts and post it raw. I have zero doubts about who's really in charge of everything I write. And I refuse to accept that I should be punished because you're not capable of doing the same.

🛡️ RAI: Real. Authored. AI.

[𒎏Wii 🏳️‍⚧️]

First of all, thanks for the long response; but the impression I get, at least if I understand your explanation correctly, is that you're barely going beyond using AI for proof reading at that point, no?

In your post, you brought up that medium doesn't allow "Anything more (drafting, editing, all other AI assistance)?", which is what I find reasonable.

There is definitely a bit of a blurry line between simply having AI correct spelling and point out segments that need to be removed and straight up re-writing a post; if that's where you want to see more nuance, then I can partly agree to that. Partly, because there is still the problem that all current AI is trained on stolen content, so any usage of AI while working on an article is still making use of another writer's work without their consent.

[Ashley Childress]

First up: the whole “all AI is trained on stolen content” claim is just... not it. That’s basically the internet equivalent of shouting “fire!” in a movie theater because you smelled popcorn. If anyone actually has a forensic report, a real dataset audit, or something besides the usual doom scroll, I’d genuinely love to see it.

Until then, it’s just rhetorical smoke—big on drama, light on receipts.

Now, to your actual question:

“You're barely going beyond using AI for proof reading at that point, no?”

Yep, you’re making my point for me. Where’s the line, anyway? Is editing, restructuring, or even just brainstorming with AI suddenly off-limits, or does it only count when you let the bot write the punchline? The definitions are fuzzy at best, so I go out of my way to make sure I’m playing by the rules—and still being transparent about how I use these tools.

My issue with Medium's post isn’t just about the “gray area” stuff—it’s how the line gets drawn. For me? Every draft starts as a mess of my own words, scattered notes, maybe an emoji or three, and not a header in sight. When the AI helps shape that chaos into something readable, Medium says it’s “AI-generated.” (And honestly? They’re not wrong.) 🤷‍♀️

But here’s the thing (and yeah, I’m biased): not all AI content is created equal. There’s a big difference between dumping two lines into a bot and calling it done versus using it to wrangle your own spaghetti draft into something real.

Intent and process matter, not just outcome. Are you using AI as a shortcut to pass off something you barely touched, or are you collaborating, rewriting, putting your own spin on it? Totally different ballgame.

My Problem With This From the Start

Lumping every AI-assisted writer in with the spammy mass-producers just means the good-faith folks get steamrolled right along with the bots.

The platform’s energy should be spent on stopping plagiarism and the endless flood of AI sludge — not on gatekeeping honest writers out of existence.

---

🛡️ This reply brought to you by a very real human, with a little help from the robots.

[𒎏Wii 🏳️‍⚧️]

AI crawlers get caught all the time. I've had to block plenty of them from my own domains. Sorry to burst your bubble, but if you're using almost any of the big AIs, you are stealing my and lots of other people's content. This is a fact.

[Ashley Childress]

You’re saying you’ve got proof that crawlers are “stealing” public data you put out on the open web? "Sorry to burst your bubble", but reading public info you freely share online is not stealing. Don’t take my word for it—see hiQ Labs v. LinkedIn where multiple courts upheld the decision.

Yeah, sometimes companies cross the line. I'm not saying it never happens so if you have actual evidence—like a legal decision, a forensic audit, or anything beyond rumors—send it my way. Otherwise, you’re just repeating Medium’s “AI is evil” panic message without receipts.

What I'm gathering from your statement is you've got a robots.txt saying “no crawlers”? Yes — they should respect it. Many don’t. But:

That’s still not theft under U.S. law.
Again, see hiQ Labs v. LinkedIn and this legal explainer

So—what’s your actual proof? I’ll happily read real sources, like:

• Court cases (hiQ Labs v. LinkedIn)
• Peer-reviewed research (A Common Pool of Privacy Problems, arXiv)
• Legal explainers (Is Web Scraping Legal?)

Until then, it’s just another myth that keeps spreading because no one stops to check the facts.

As for me? My robots.txt is a giant "come on in" sign. I want crawlers to use my stuff. More reach, more impact. Why? AI isn’t the villain—people are the ones who choose to do harm.

---

🛡️ Written by me & the robots

[𒎏Wii 🏳️‍⚧️]

As for me? My robots.txt is a giant "come on in" sign. I want crawlers to use my stuff.

Cool, it's your right to consent to things.

It is also my right to not consent, and this consent is routinely violated by AI companies. Refusing to admit this means you're refusing to consider the moral implications of your own decisions, and that is simply wrong.

The fact that you bring up the US legal system to support your views... I don't need to post a list of extremely upsetting facts here. We all know how fucked up that is as an argument.

Side note: I didn't bother bringing this up because it doesn't affect me directly, but AI companies have also been caught DDoSing a large number of free services like Wikipedia, etc.

For anyone reading along: You can use Anubis to protect your content from these attacks, both to prevent outages and to protect your intellectual property.

[𒎏Wii 🏳️‍⚧️]

PS:

diff.wikimedia.org/2025/04/01/how-...

[Ashley Childress]

Let’s reset for a second 🛑

I’m starting to feel preached at, not debated with. I don’t think we’re miles apart, but some of the way you’re phrasing things is working against you — and I want to show you why.

“consent is routinely violated by AI companies”

“refusing to admit this”

The problem isn’t the concern — I’ve already said “sometimes companies cross the line”. That’s fair. The issue is the word theft. When you jump straight to that, it collapses all nuance. If you instead framed it as:

👉 “companies routinely overstep posted consent signals like robots.txt”

—you’d keep the moral weight without overstating the legal claim. That’s stronger.

---

On law and standards ⚖️

“The fact that you bring up the US legal system to support your views...”

I only referenced U.S. law because it’s the framework I live under — it’s the well I can draw facts from. That doesn’t mean I think it’s flawless. If you want to broaden it, cite examples from EU copyright directives or Canadian fair dealing law. That would actually make your point land harder because it shows the pattern across borders.

---

On asides and evidence 📝

“I didn’t bother bringing this up because it doesn’t affect me directly...”

This is where you undercut yourself. If it doesn’t affect you, it reads like an aside, not evidence. But if you have credible examples — DDoS logs, published cases, reports — lead with those. Don’t bury them as a side note. That way it feels like proof, not padding.

---

On your “PS” 📎

As for your "PS"

diff.wikimedia.org/2025/04/01/how-...

This is a great example. It actually shows cause. It does absolutely nothing to back up your theft claim. But it does show harm that is quantitative.

However, what Wikimedia is saying is not that they’re stealing content — it’s that they’re not being reimbursed for the costs they incur by supporting scraping traffic on their site. Not the scraping itself, but the load they attribute to scraping.

And they back that up clearly with one line from your source:

“Our content is free, our infrastructure is not.”

So the real problem is: how do you differentiate? You either charge everyone or no one. Where’s the middle ground? How do we fix it? I don’t know — but those are the right questions to be asking.

---

On blanket claims 🚫

“If you don’t write your own content, you aren’t a writer.”

“If you’re using almost any big AI, you are stealing.”

Both of these are blanket statements. Blanket statements sound punchy, but they’re easy to dismantle because they sweep too wide. Try tightening them into something like:

👉 “AI makes it easier than ever to pass off generated work as original.”

Now the focus is on behavior, not on tarring an entire group with one brush.

---

Where I stand vs. where you could strengthen 💡

• Me: I’ve explained my workflow enough to show intent and craft. I’ve been clear that bad actors exist but “all AI = stolen” doesn’t pass the evidence test.
• You: Avoid blanket terms, avoid undercutting asides, ground in evidence (or multiple jurisdictions if you want extra punch).

Because in the end, if the goal is protecting both readers and creators, the way forward is pretty simple: target behavior, not the tool.

✅ Punish plagiarism and deceit.

✅ Punish lazy mass-generation.

✨ Reward transparency and craft.

We’ll get further if we build solutions instead of lazy gatekeeping.

---

🛡️ RAI: Responsible AI in practice means disclosure, craft, and intent — not shortcuts. That’s the standard I hold myself to here.

[Martin Falk Johansson]

Sounds like a very nice rule - I wish Dev.to had it. Since ChatGPT this place has become nigh useless since it's over 90 % AI slop - and even things that are not made by ChatGPT now somehow imitates how ChatCPT would make it.

I really don't want to bother reading something that someone did not bother writing. We're approaching the dead internet were we're wasting computer power generating text with AI that will then be scraped summarised ny another AI. Yuck.

[Ashley Childress]

I definitely encourage you to read the reply above too—it covers most of what I’d say. But I’ll add this:

Once again, I’m not actually disagreeing with you! But here’s the kicker: this whole post and the comment above? Yep, both AI-generated. So, painting all AI output as “just slop” feels a bit like… well, like saying all pizza is cardboard just because you once had a bad slice. 🍕

What really gets under my skin? It’s that knee-jerk urge to toss out anything new—just because it’s unfamiliar, misunderstood, or the loudest person in the room misused it first. And let’s be real: we don’t even have guardrails in place yet!

I’m all for setting healthy boundaries with this stuff... but you can bet I’ll speak up for myself (and my work) every single time this debate pops up.

[Colin Principe]

As someone with a degree in English Literature, and for whom effective writing has been a key part of my career even though I have not been a professional writer, I land somewhere between Medium's definition and yours.

My chief experience with AI thus far has been employing it to conduct focused information searches, provide outlines for non-work-related projects, customize resumes and cover letters in my job search, and help with refactoring code on a hobby project.

You may think this old school, but I think for a piece to have a unique voice, the writer needs to physically live with it for some time. Truman Capote infamously criticised the books of the beat writers like Jack Kerouac by saying "That's not writing, that's just typing," saying that vomiting words out without any reflection doesn't result in anything valuable. (I happen to like the Beats and disagree with Capote but I get the gist of what he was trying to say.)

Editing is a huge pain in the ass and I would love for AI to do it for me. The only problem is that often during editing I discover that I want to use a different word, express a thought a slightly different way, or restructure an argument. AI isn't going to do that for me.

In your Medium post you talk about how much effort you put into using AI to polish your content once you've done a bunch of thinking. If I could be conceited for a moment, when I write I do all that, except I do that all inside my head, based on a lifetime of doing so. I believe I have a talent for doing so effectively and consistently, similar to how a trained chef can produce a high-quality meal quicker and more consistently than a home cook.

One thing every writer knows is that every platform has an editorial policy and none is obligated to accept your work. If Medium has made this choice, there are plenty of other platforms who are happy to accept material regardless of its origin that will allow you to build an audience.

[Ashley Childress]

Thanks for your very well thought out response — I really appreciate your input 🙏

You said:

"You may think this old school, but I think for a piece to have a unique voice, the writer needs to physically live with it for some time."

I actually agree with that — and I do let AI take my draft and help organize my thoughts into something easier to read and digest. But that’s very different from not living with it. My editing process usually takes hours — sometimes even days — before I’m satisfied ⏳.

That’s exactly why your phrasing resonated with me:

"I discover that I want to use a different word, express a thought a slightly different way, or restructure an argument."

That’s what the process looks like for me too. I’ll re-read, realize a section has no oomph ⚡, and then I’ll prompt AI with something like “this isn’t working — here’s the rambling, colorful explanation of what I mean”. It’ll give me four to six options to choose from. Sometimes one works, other times I mash them together 🧩 or push it further — but it’s still my editing.

The truth is, I actually enjoy editing. I just also enjoy the tech 🤖. Writing with AI is simply more fun for me than writing without it, and that’s how I choose to make it work.

As for Medium, I was never chasing notoriety. What bothers me isn’t the lack of visibility, it’s the blanket policy 🚫. It stereotypes an entire group of writers because of a few bad actors. I’m not saying “let all AI content through” — that would have the opposite effect. What I’d love to see is something like a pre-approval badge 🎟️: your AI-supported writing gets personally reviewed once, approved for their programs, and then treated as normal unless it breaks the rules later. That way Medium still has quality control, but writers aren’t punished up front just for using modern tools.

That’s the middle ground that’s missing right now — the part where trust is earned instead of assumed lost 🤝.

[Anik Sikder]

This hits hard. Medium’s stance feels like it was written by someone who’s never actually used AI creatively. You’re right, it’s not about outsourcing your voice, it’s about sharpening it. Blanket bans don’t protect quality, they just silence nuance. Glad you spoke up.