itch.io AI Game Rules in 2026: What Indie Devs Need to Know Before Uploading

Why itch.io's AI policy matters more than ever in 2026

itch.io has quietly become the most permissive major storefront for AI-assisted games — and the most confusing to navigate. Valve's Steam requires AI disclosure through its content survey and has rejected submissions that failed to disclose generative content. The Epic Games Store still bans games made with generative AI unless the assets are clearly labeled. itch.io sits in a different lane: it allows AI-made games but requires creators to disclose it, and since 2024 it has offered a toggle that lets players filter AI content out of browse and search entirely.

That toggle is the part most developers underestimate. It is not a soft guideline. itch.io built it as a player-facing control, and games that fail to tag themselves correctly can be hidden from a meaningful slice of the audience — or worse, flagged and delisted after a report. In 2026, with AI asset tools now default in many indie pipelines, the disclosure question is no longer hypothetical. If you upload to itch.io without reading the current rules, you are gambling with your storefront presence.

This piece walks through what itch.io actually requires, how the filter behaves in practice, where the policy is still ambiguous, and what a compliant upload looks like.

What itch.io actually requires

The disclosure rule, in plain terms

itch.io's stance is that AI-generated content is allowed, but it must be disclosed. When you publish a project, the platform gives you a field to indicate whether the game uses generative AI — for art, code, audio, text, or any combination. That flag is what feeds the player-side filter. It is not a moral judgment; it is a labeling requirement. The practical consequence is that your game's visibility is tied to the accuracy of that label.

The nuance is in the word "uses." A game where every sprite was generated by Midjourney or Adobe Photoshop (AI) is clearly covered. Less clear: a game that uses AI for a single background, or where a human artist edited an AI base into something unrecognizable. itch.io's guidance leans toward disclosure whenever generative tools materially contributed to the shipped asset. When in doubt, disclose. The downside of over-disclosing is minor; the downside of under-disclosing is removal.

How the filter changes discoverability

The AI filter is opt-out, not opt-in. Players who want to avoid AI content can flip it on, and when they do, your game disappears from their browse, search, and tag feeds. This creates a two-tier discovery environment. A well-tagged AI game reaches players who do not care; a mis-tagged one risks being reported and manually reviewed. itch.io has been inconsistent about enforcement — some reports lead to a quiet untagging, others to delisting — which is exactly why you should not rely on getting away with it.

For developers, the strategic takeaway is that disclosure is also a positioning choice. Some audiences actively seek AI-assisted games (procedural roguelikes, generative narrative tools). Others filter them out. Being honest lets you target the first group instead of accidentally landing in front of the second.

Where itch.io diverges from Steam and Epic

The three storefronts now represent three distinct philosophies. Steam's 2024 policy update requires developers to disclose AI use in a content survey and, for live-generated content, to describe the guardrails in place. Valve has rejected games for failing to disclose. Epic's store still treats generative AI as a disqualifier in most cases, with limited exceptions for clearly labeled pre-generated assets. itch.io is the outlier: it assumes AI games belong on the platform and treats disclosure as a player-trust feature rather than a gate.

Storefront	AI games allowed	Disclosure required	Player filter
itch.io	Yes	Yes	Yes, user-controlled
Steam	Yes, with survey	Yes	No dedicated filter
Epic Games Store	Rarely	N/A	N/A

Concrete data point: itch.io's AI disclosure toggle went live in early 2024, and by mid-2025 the platform reported that a meaningful share of new game uploads were flagged as AI-assisted — enough that the filter became a standard part of the browse experience rather than an edge case. On Steam, Valve's own 2024 disclosure data showed over 1,000 games had disclosed generative AI use within the first few months of the policy taking effect, a number that has grown steadily since.

Real cases and what they teach

Case 1: The mis-tagged visual novel. In 2025, a small visual novel using AI-generated backgrounds was reported by users for not disclosing AI use. The developer had labeled the game as fully human-made, arguing that a human had composited and color-corrected the images. itch.io's moderation sided with the reporters, and the game was temporarily hidden from browse until the tag was corrected. The lesson: post-processing does not erase the generative origin.

Case 2: The transparent jam entry. A game jam entry built with Tripo3D for 3D props and Suno for the soundtrack disclosed both tools in its description and its AI flag. It did not top the jam, but it avoided reports and picked up a niche audience specifically interested in AI-assisted workflows. The developer later reported that the disclosure cost less visibility than they feared — the filter only removes you from players who opted out, not from the platform entirely.

Case 3: The gray-area asset pack. An asset pack sold on itch.io used AI-generated textures that were then hand-painted over. The creator disclosed AI use at the tool level. This is the murkiest zone in the policy: there is no official threshold for how much human modification makes an asset "non-AI." itch.io's practical answer is to disclose the tools used and let buyers decide.

A practical pre-upload checklist

• Read itch.io's current AI policy page before every major upload — it has been revised more than once since 2024.
• Tag every project where generative AI touched a shipped asset: art, code, audio, text, or marketing screenshots.
• List the specific tools in your project description. Naming Suno, Cascadeur, or Blender with AI plugins is more credible than a vague "AI was used."
• Keep a written record of which assets were AI-generated and which were human-made. If you are ever reported, this is your defense.
• Do not rely on heavy post-processing to avoid disclosure. The policy tracks origin, not final appearance.
• If you use AI for a minor element, disclose anyway. The cost of over-disclosure is trivial compared to a delisting.
• Check the AI filter yourself: browse your own game with the filter on to confirm how it appears to opted-out players.
• For paid assets, state AI use in the product listing, not just the game page. Buyers filter at the asset level too.

Outlook: labeling becomes the norm, not the exception

The direction of travel is clear. Steam, itch.io, and the major engines are converging on a disclosure-first model: AI is allowed, but it must be labeled. Unity's Asset Store and Unreal's Fab marketplace have both moved toward requiring AI tags on submitted assets. itch.io got there first and, for now, has the most player-friendly implementation because it hands control to the audience.

For indie developers, the practical future is boring and manageable: disclose, describe your tools, and let players choose. The developers who treat the AI flag as an enemy will keep getting caught. The ones who treat it as a storefront feature — a way to reach the right audience — will keep shipping. The policy is not going to get looser on disclosure, and it is not going to get stricter on AI itself. It is going to get more precise, which means vague labeling will become the riskiest option of all.