Nexus Mods makes AI disclosure mandatory: new mods declare at upload, existing ones have until January 2, 2027

Nexus Mods, the largest modding site for PC games, made AI disclosure mandatory on November 3, 2026: every newly uploaded mod must now declare whether generative AI was used, and existing mods have a 60-day grace period that ends January 2, 2027. After that date, non-compliant mods go to the site’s Trust & Safety team. Site owner Victor “Foledinho” Folmann framed the policy as pro-human creativity — but the site chose disclosure over a ban.

Key points

• New uploads have had to declare AI use at upload time since November 3, 2026; mods tagged AI-assisted must also publish a public explanation of how and where AI was used.
• Existing mods get 60 days, until January 2, 2027; editing an older mod now shows a persistent reminder in the upload form, with a dismissible popup on My Mods.
• Nexus Mods pledges not to train AI on mods and not to sell or license them to anyone who would; AI-training and forking permissions default to “Ask Me”, covering legacy mods too.

Background

The AI tags themselves are not new. Nexus Mods has carried voluntary AI-generated and AI-assisted tags for a while, and it already updated its generative-AI tagging and moderation guidelines once, in late July (per Shacknews). What changed is the enforcement: declaration went from optional to a hard gate on the upload form. The pressure shows in the site’s own numbers, published in the announcement: between November 2025 and October 2026, staff manually reviewed 3,164 uploads, more than four times the previous year, and quarantined about 20,000 uploads — twelve times the prior year. Of this year’s Donation Points, 2.13% went to mods tagged AI. Community feedback, the site says, kept returning to two complaints: undisclosed AI use, and authors dumping mods in batches that flood the Recently Added page. Platform-level AI governance in gaming has been tightening for weeks — Anthropic’s server-side rule against game recreation in Claude Code stirred its own controversy in late October — but Nexus Mods took a different route: no ban, full disclosure.

The rules

The policy splits disclosure into tiers, and the deadlines differ between new and existing uploads:

Requirement	New uploads	Existing mods
AI declaration	Mandatory since 2026-11-03	Backfill by 2027-01-02, then Trust & Safety
AI-assisted extra duty	Public explanation of tools and use	Same; persistent form reminder on edit
Display	AI tags shown right after adult-content tags	Players can hide AI-generated content on game pages and the homepage

AI-generated covers mods whose core work came out of an AI tool — generated code, voice lines, textures, text or scripts. AI-assisted covers human-led work where AI helped (texture upscaling, audio cleanup, initial concept art), and that tag raises the evidentiary bar: the help center requires a public statement of which tools and models were used and which outputs ended up in the final files, plus development-history evidence — version control commits, project files, prior work. Mods that misuse the tag can have it replaced by moderators. AI outside the game files — promotional images, thumbnails, banners, videos, description text — falls under disclosure too. On the player side, hide controls for AI-generated content are live on game pages and the homepage; search filtering is “next up”, per The Verge. Against batch uploads, the site is capping how many mods from one author show in Recently Added and tightening upload restrictions on newly created accounts, with details “coming soon”. The consequences are spelled out: after a staff notification, continued failure to update a mod leads to moderation action, account restrictions, warnings or removal.

What others say

Folmann’s case against a ban: it “will not eliminate AI from the modding process but will prevent people from mentioning its use.” He did not present the policy as final — the rules “will need to evolve”, and “it’s definitely not a conversation we’re going to have once” (as VGC quoted him). The no-training pledge is categorical: “We refused to do that in the past, and we will refuse to do so in the future.”

The community split is visible under VGC’s report: one side wants an outright ban and complains AI thumbnails are taking over mod pages; the other says AI-assisted coding lowered the barrier to modding and feels targeted by the new rules. The industry is taking three roads at once. Itch.io, the same day, said it had deleted “a couple of terabytes of AI slop” and rolled out generative-AI disclosure filtering. Valve is pushing review times back down as AI-assisted submissions flood Steam — delays reached 12 business days against a 3–5 day target, per TechRaptor. Epic’s Tim Sweeney, meanwhile, stands publicly behind generative AI content.

Our take

For mod authors the order of operations is simple. If AI was used, backfilling the declaration inside the grace period beats being reported after January 2. Before choosing AI-assisted, check you can actually produce the evidence — version history plus a tool list is a real burden for a solo author. And “no AI used” deserves a check of its own: a description written by AI counts as AI media under the policy.

Enforcement is the soft part: declaration relies on authors, discovery on player reports, review on staff. Until January 2, most existing mods carry no AI tags at all, so the player-facing filters are only as good as the disclosure rate. The transferable lesson for anyone building a content platform is the permissions design: Nexus Mods made “may my work train AI” an explicit default — “Ask Me” — covering legacy content, the same direction as the AI training opt-out token Anthropic pushed into the IETF: preferences written as explicit declarations, this time shipped as a product default. Three things to watch: the search filter launch, the promised upload-limit details, and the first removals after January 2, 2027.