itch.io AI Game Policies in 2026: A Transparency-First Guide for Indie Devs

Why itch.io's AI Rules Matter Now

itch.io has become the default storefront for experimental, small-team, and AI-assisted games. That is precisely why its AI policy matters more than most developers realize. Unlike Steam, which since January 2024 has required developers to disclose AI use at submission and has rejected a small but growing number of titles for undisclosed generative content, itch.io has taken a lighter-touch, community-trust approach. The platform's guidelines, updated through 2025 and visible on its creator documentation pages, require sellers to disclose AI-generated assets and to avoid flooding the marketplace with low-effort, mass-produced content. Enforcement is reactive: reports, manual review, and account-level penalties rather than automated pre-screening.

The practical consequence is that compliance is mostly on you. A game that ships with undisclosed Midjourney backgrounds or Suno-composed tracks may stay live for months, but it is one report away from delisting, and repeated issues can affect your account's standing. In 2026, with AI asset tooling now embedded in nearly every creative pipeline, the question is no longer whether you use AI, but whether your store page tells the truth about it.

What the Policy Actually Requires

Disclosure, not prohibition

itch.io does not ban AI-made games. Its rules focus on two things: transparency about AI-generated content, and quality control against spam. In practice, that means your project page should state which parts of the game used generative AI, and your uploads should not be bulk-generated variations of the same asset pack. The platform's terms give it discretion to remove content that misleads buyers or degrades the browsing experience.

This is a meaningful difference from Steam's more formal disclosure system, where developers must describe AI use in a submission form and Valve reviews it. itch.io's model is closer to a community norm: it works if creators act in good faith, and it fails when they do not. For indie developers, the safest interpretation is to err toward more disclosure. A short "AI disclosure" section on your page costs nothing and protects you if a buyer or moderator questions an asset.

The gray zone: tools versus generation

The policy gets murky around tools that contain AI features but are not purely generative. If you use Photoshop's Firefly features to touch up a hand-painted sprite, is that AI-generated content? Most reasonable readings say no, provided the core asset is human-made and the AI is an editing aid. But if you generate a full background from a text prompt, that is clearly generative and should be disclosed. The line is authorship: did a human make the substantive creative decisions, or did the model?

This distinction matters for 3D and audio too. A character modeled by hand in Blender but textured with an AI-generated PBR set sits in a gray area. A track composed in AIVA from a prompt is straightforwardly AI-generated. When in doubt, disclose and describe the workflow. Buyers respond well to specificity, and moderators are far more likely to leave a transparent page alone.

Real Cases and Data Points

Steam's approach provides the clearest comparison. Valve began requiring AI disclosure in January 2024 and by mid-2025 had rejected or delisted a number of titles for failing to disclose generative content, according to developer reports compiled on Steam's community forums. The company also clarified that live-generated content, such as an LLM-driven NPC, requires additional safeguards and disclosure. itch.io has not published equivalent enforcement statistics, which is itself a signal: the platform prefers case-by-case moderation over public rulemaking.

On the tooling side, the market has matured fast. Scenario launched its game-asset generation platform in 2023 and by 2024 was positioning itself around style-consistent, rights-cleared asset pipelines for studios. Unity's Asset Store and Unreal's Fab marketplace have both introduced AI content labeling requirements, with Fab launching in October 2024 and explicitly asking sellers to flag AI-generated assets. These moves normalize disclosure across the ecosystem, which in turn makes itch.io's lighter rules easier to follow: if you are already labeling assets for Fab or the Unity Asset Store, you have most of the information itch.io expects.

One more data point worth noting: itch.io's own game jam culture. Jams like the annual GMTK Game Jam, which drew over 7,000 entries in 2024, have become testing grounds for AI-assisted workflows, and jam organizers increasingly ask participants to disclose AI use. That grassroots norm is likely shaping itch.io's moderation posture more than any formal policy update.

Practical Advice for Developers

• Add an AI disclosure section to every project page. List tools used, which assets were generated, and whether any content is generated at runtime. Keep it short and factual.
• Separate generative from assistive use. Hand-painted art refined with AI filters is different from prompt-to-image generation. Describe the difference explicitly.
• Avoid bulk uploads. itch.io's spam rules target mass-produced asset packs and low-effort variants. Ship fewer, more distinct projects.
• Keep a tool log. Track which assets came from which tool and version, including Tripo3D or Suno. This makes disclosure easy and protects you if a claim is challenged.
• Check licenses before selling. Some AI tools grant commercial rights only on paid tiers. Verify before you list a game for sale.
• Respond to reports quickly. If a moderator flags your page, a prompt, honest reply with updated disclosure usually resolves the issue faster than arguing.

Outlook

The direction of travel is toward more formal disclosure, not less. Steam's enforcement has already set a precedent, and marketplace labeling from Fab and the Unity Asset Store is making AI provenance a standard metadata field. itch.io is likely to follow with clearer guidance rather than a hard ban, because its community depends on AI-assisted creators and its moderation capacity is limited. Developers who build disclosure into their workflow now will be ahead of the curve; those who treat it as an afterthought will keep getting caught by reports and policy updates. The safe bet for 2026 is simple: be specific about what you made and what the model made.