itch.io AI Game Policies in 2026: A Store-Page and Asset Strategy Guide

Why itch.io's AI Rules Matter More in 2026

itch.io has never been Steam. It's the platform where a solo developer can upload a prototype at 2 a.m. and have a page live by 2:05. That openness is exactly why its AI policy matters: itch.io is the default home for experimental, AI-assisted indie work, and its rules shape what small teams can ship without friction.

The pressure has been building from both sides. Steam's AI disclosure system, introduced in January 2024, forced every storefront to articulate a position. itch.io responded with its own approach: a disclosure requirement and a ban on certain AI-generated assets, enforced through a mix of automated flags and manual review. By 2026, the practical effect is visible in how developers structure their pages, not just in what tools they use.

The problem this article solves is narrow but real: developers know they need to disclose AI use, but they don't know what actually triggers rejection, how much detail the review process expects, or how to present AI-assisted assets without hurting their page's conversion. The gain is a repeatable store-page and asset strategy that survives review and still reads well to players.

What the Policy Actually Requires

Disclosure is the baseline, not the whole story

itch.io's core requirement is disclosure: if a game contains AI-generated content, the developer must say so on the project page. That sounds simple until you hit the edge cases. Does a procedurally generated texture count? A voice line run through an AI cleanup tool? A code snippet from a coding assistant?

The platform's working distinction is between AI as a production aid and AI as a content source. Tools that assist human work — an AI upscaler, a code completion model, a noise-reduction pass on recorded audio — generally fall into a gray zone where disclosure is recommended but not always mandatory. Assets generated wholesale by a model — character art, music tracks, voice lines, 3D meshes — are the ones that require explicit disclosure and, in some categories, are restricted outright.

The asset bans are narrower than developers fear

The phrase "asset bans" gets repeated in developer forums with more alarm than accuracy. itch.io has not banned AI-assisted development. What it restricts is the sale or distribution of certain fully AI-generated assets, particularly where the platform cannot verify provenance or where the output competes directly with human-made asset packs. The enforcement tends to focus on store pages that present AI-generated art as hand-drawn, or asset packs that are clearly model output with no human editing pass.

The practical takeaway: a game with AI-assisted background art and human-designed characters, disclosed clearly, is in a very different position from a page selling a pack of unedited Midjourney outputs. The first is routine. The second is where rejections cluster.

How Review Actually Works in Practice

Triggers and the manual review queue

itch.io does not publish a scoring rubric, but developer reports through 2024–2025 point to a consistent pattern. Pages get flagged when the description contains certain phrases ("AI-generated," "made with Midjourney," "AI art"), when uploaded assets match known model outputs, or when a user report is filed. Flagged pages go to manual review, which can take days and, in peak periods, longer.

The most common rejection cause is not the presence of AI content. It's inconsistency: a page that discloses AI in one sentence and then describes the art as "hand-painted" in the next. Reviewers read the whole page. If the disclosure contradicts the marketing copy, the page gets bounced.

Store-page structure that passes review

A compliant page has three things: a clear disclosure section, a consistent asset description, and a contact path. The disclosure doesn't need to be buried in a terms-of-service block. A short section near the bottom of the description — "AI disclosure: background textures generated with [tool], then edited in [tool]" — is usually sufficient and reads as transparency rather than confession.

Consistency matters more than length. If your capsule art is AI-assisted, say so where you describe the art. If your soundtrack uses an AI composition tool like Soundraw, name it in the credits and the disclosure. The goal is that a reviewer can trace every AI touchpoint without hunting.

Real Cases and Data Points

The clearest reference point remains Steam's January 2024 disclosure rollout, which required developers to describe AI use at submission and made that disclosure visible on store pages. itch.io's approach is lighter-touch but follows the same logic: disclosure first, restriction second. Developers who shipped on both platforms in 2024 reported that itch.io review was faster but less predictable, because it relies more on manual judgment than a structured form.

On the tooling side, the asset pipeline has matured enough that disclosure is now a documentation problem rather than a production problem. Tripo3D and similar text-to-3D tools can generate a prop mesh in minutes, but the game-ready version still needs retopology and UV work in Blender. That human pass is exactly what distinguishes a disclosable AI-assisted asset from a banned raw output. The same pattern holds for 2D: a concept generated in an image model and then repainted in Krita has a documented human editing step that a raw generation does not.

A second data point worth noting: itch.io's own game jam ecosystem. Jams with AI-permitted rules have grown steadily since 2023, and jam organizers increasingly require disclosure in the submission form. That has normalized the practice at the community level, which in turn has made reviewers more comfortable with disclosed AI use on regular pages. The direction of travel is toward documentation, not prohibition.

Practical Advice for Developers

• Write the disclosure before you write the marketing copy. Draft the AI section first, then make sure every claim elsewhere on the page is consistent with it.
• Name the tools and the human editing step. "Generated with X, edited in Y" is stronger than "contains AI content" because it shows a production process.
• Keep raw model output out of asset packs. If you sell assets, the human pass is the difference between a listing and a rejection.
• Separate AI-assisted from AI-generated in your own mind. Upscaling, cleanup, and code assistance are low-risk; wholesale generation of art, music, or voice is where disclosure must be explicit.
• Screenshot your pipeline. Save layer files, project files, and edit history. If a page is flagged, provenance evidence resolves the review faster than any appeal text.
• Check jam rules separately. A compliant store page does not guarantee jam acceptance; organizers set their own disclosure requirements.

Outlook

The trajectory is toward structured disclosure rather than broader bans. Steam's form-based approach and itch.io's manual-review approach are converging on the same outcome: AI use is permitted when it is documented and when a human production step is visible. Developers who treat disclosure as part of the store page — not an afterthought — will find the process routine. The ones who get rejected will be the ones whose pages contradict themselves.

The bigger shift is cultural. As AI-assisted pipelines become standard in small-team production, the question is shifting from "did you use AI" to "can you show your work." itch.io's policy, for all its ambiguity, is already built around that question.