We’ve done this before
I’ve seen this movie. It came out when Photoshop got good.
Remember that? “It’s not real art if you used layers.” “Real photographers don’t retouch.” People wanted labels on edited images, purity tests for digital artists, the whole thing. And then… everyone used Photoshop. Nobody labels it now, except in the places where faking it actually hurts somebody, like news photos and contest entries. The tool stopped being the question and we went back to judging the work.
So when Nexus Mods dropped its big AI policy post on November 3rd, and the comment section turned into a 200+ comment cage match within hours, I had a little deja vu.
My comment was simple. People are going to write bad mods no matter what tool they use. You shouldn’t punish people for using AI just because they can do it faster. Bad mods existed long before ChatGPT. Plenty of broken, half-finished junk got lovingly hand-made by humans, slowly, the old-fashioned way.
Credit where it’s due
I’ll give Nexus this much: they did not do the dumb thing.
There’s no outright ban. Donation Points aren’t changing (AI-tagged mods pulled 2.13% of all points this year, which they called a comfortably low share). And they actually said the quiet part out loud: if you punish people for disclosing AI, you just teach them to hide it. Correct!
The rules with actual teeth are aimed at behavior, not tools:
- Using someone else’s mod without permission? Not hosted. Doesn’t matter how you made it.
- Fake “screenshots” showing features your mod doesn’t have? Against the rules, AI or not.
- Malware? They quarantined 20,000 uploads for extra security checks over the last year, 12 times the year before, and they’ll ask for source code when something smells off.
- New AI-training and fork permissions that default to “Ask me,” so authors get a say before their work gets reused.
They also promised they won’t use your mods to train AI or sell them to anyone who will. Good. That’s the stuff that matters.
So why am I mad?
Where’s the line, though?
Here’s my actual question. Where is the line?
Because Nexus doesn’t know either. That’s literally their reason for not banning AI. Their own post says it’s hard to draw “a clear boundary between what is acceptable and unacceptable” when AI shows up everywhere from script autocomplete to full-on generated textures and voices.
Cool. Agreed. So what did they ship?
A checkbox. A yes/no box. Mandatory on every new mod, starting today, with a 60-day grace period for existing mods that ends January 2nd, 2027. After that, skip it and you’re looking at warnings, restrictions, or removal.
They said there’s no clean line, then built a feature that pretends there is one.
And think about who that box actually catches. The person who used Copilot to finish a C# patch ticks it. The person who generated every texture and then “edited them a little” maybe doesn’t. The person actually trying to scam people definitely doesn’t, because liars lie on checkboxes. That’s kind of their whole thing.
To be fair, there’s also a required “explain how AI was used” field for AI-assisted mods, and honestly? That’s the only part that makes sense. “Upscaled one texture” and “the entire mod is Midjourney” are not the same thing, and a sentence of context tells you that. The checkbox throws them in the same bucket.
And if your line is truly zero AI, anywhere, ever… I don’t know what to tell you. Autocomplete, upscalers, translation tools, half the games you’re modding. If you don’t want any AI anywhere near your mods, you honestly shouldn’t use mods.
Why is it sitting next to porn?
This is the part that genuinely gets me.
The AI tags now show up right after the adult-content tags on mod pages, and the content filter toggles live right below the NSFW ones. Placement is a statement. It tells you what category something belongs in. And sitting next to porn says: this is objectionable, shield your eyes.
The NSFW filter exists because some content isn’t appropriate for some audiences. A kid shouldn’t stumble into it. Fine. Makes sense.
An AI-upscaled texture isn’t that. Nobody is harmed by seeing a mod where someone used autocomplete. It’s a taste thing, same as “I only want lore-friendly armor” or “hide mods that need SKSE.” Put it with the preference filters, where it belongs, and nobody reads it as a moral judgment.
Instead, honest authors get a scarlet letter in the most loaded spot on the page. And the person who lied? Clean page. No tag. Looks great.
That’s backwards.
The comment section, though
I went through the comments (219 and counting when I looked) and wow. There are some hot takes in there. But there’s also a lot of people saying the thing better than Nexus did.
ZArkadia nailed it in like two lines:
“Disclosure is not the issue. The attacking and harassment after disclosing is the issue.”
That’s it. That’s the whole problem with a scarlet-letter checkbox. SIIIN3YUM1, who opened the thread saying they’re happy to disclose exactly how they use AI, was basically agreeing in real time.
utherix went straight for the code tag, pointing out that most code written in 2026 comes out of an LLM anyway, and asked why we’d label that when
“You wouldn’t have a tag that says ‘Code written using an IDE’.”
Exactly. duffbeeeer wanted the same fix from the other direction: split the tag into subcategories, because AI-assisted code and full generative art don’t belong in one bucket.
Squibster99 brought up the part nobody else wants to talk about. They use AI for all their mods because of their technical skills and health, and they’d struggle to mod at all otherwise. They still support disclosure and filtering. That’s what accessibility looks like, and it’s getting lumped in with spam.
Now, to be fair to the other side (I’m not going to pretend it doesn’t exist):
- TadasTheLithuanian described the Bannerlord section going from a couple new mods a day to a full page of mostly AI stuff every 24 hours. That’s real, and it sucks for human modders trying to get seen.
- GoblinHours69 pushed back on the “people who can’t code” argument, saying anyone can learn with time and discipline, and that LLMs are actually decent teachers now.
But look at those complaints. Flooding is a volume problem, and Nexus is already capping how many mods one author can push into Recently Added. Learning to code is a personal choice. Neither one gets fixed by a checkbox.
And WyattTheSans called the whole vibe of the thread before it even started: entitlement from both sides and the fence-sitters. Yeah. That’s the internet.
Bad people, not bad tools
Stealing someone’s mod is bad people doing bad people things. Faking screenshots is bad people doing bad people things. Hiding malware in a texture pack? You guessed it.
AI makes some of that cheaper and faster. Sure. So did the internet. So did email, which is why we have spam filters, and nobody decided email itself was evil and slapped a warning label on every message.
Nexus already has the right tools in the post. Permissions. Security reviews. Rules against misrepresenting your mod. Upload caps. Go after the behavior, hard. Keep the filter if people want it, but move it out of the porn neighborhood, make the “how I used AI” note the main thing instead of a yes/no box, and stop treating honest authors like the problem.
Give it five years. “AI-assisted” on a mod page is going to read exactly like “made with Photoshop” does now.
Technically true. Nobody cares.
Sources: AI-Generated Content on Nexus Mods (Nexus Mods news post and comments, November 3, 2026)
About The Author
