Itch.io’s New ‘No AI’ Filter Gives Players a Way to Shrink the Discovery Firehose

Itch.io has added a “No AI” filter to its browsing menu, search tools and other areas where people discover projects. It is a small control with a very large visible effect: when applied to games, the selection shown drops from more than 1.77 million projects to fewer than 425,000.

That number is the immediate story. A storefront filter does not judge whether a game is good, and it does not settle every argument about creative tools. What it does is give players a direct way to state a preference while navigating a vast, open catalog. On a platform where finding an unusual little game can be the whole appeal, the ability to remove disclosed generative-AI work from results may be as consequential as a genre, tag or price filter.

What the No AI filter does

The new option is designed to exclude projects marked with generative-AI disclosures. It is available while browsing games and has also been introduced to search and other discovery surfaces that previously lacked it.

That distinction matters. Discovery surfaces are the places that determine what a visitor sees before they have chosen a specific game: browse pages, search results and similar listings. Adding a filter only to one menu would limit its usefulness. Making it available across multiple ways of finding projects gives users a more consistent choice, whether they arrive with a specific query or simply want to wander through new releases.

The feature is not presented as a claim that every project using AI is identical, nor is it an automated verdict on a game’s artistic merit. It is a preference tool: users who do not want disclosed generative-AI material in their results can reduce it, while users who do not mind it can leave the setting off.

The size of the catalog shift is hard to ignore

Moving from over 1.77 million projects to under 425,000 means the filtered result set is less than one quarter of the original total. Put differently, the No AI choice removes well over 1.3 million projects from that particular game listing.

Those figures should be read carefully. They describe the results shown when the filter is used, not a quality ranking, sales measure, or proof that every excluded project was wholly made by automated systems. They do, however, illustrate the practical reason a filter exists. At catalog scale, even players who are merely trying to locate a few human-made experimental games, game-jam oddities or small personal projects can face an overwhelming amount of material.

Generative AI generally refers to tools that create new output—such as text, images, audio or code—from prompts and other inputs. In a game listing, a disclosure can potentially concern different parts of a project rather than one uniform production process. A player may care intensely about that distinction, or may simply prefer not to parse it project by project. The filter accommodates the latter choice without requiring users to inspect every page manually.

Disclosure is useful, but it is also the pressure point

The system depends on disclosures. Itch.io already asks creators to disclose generative-AI use where appropriate, and the platform has a policy against uploading excessive amounts of automatically generated or AI-generated content, treating that behavior as spam.

Those are related but separate ideas. A disclosure policy asks creators to identify relevant use of a tool. An anti-spam policy addresses volume and behavior: dumping large quantities of automatically generated material can make a marketplace harder to browse regardless of whether individual listings technically contain a disclosure.

The new No AI filter makes accurate labeling more consequential. If a creator properly identifies relevant generative-AI use, people who want to avoid such projects can do so. If a project that should be labeled is not labeled, it may still appear in filtered results. The available policy information does not establish a stated penalty for failing to use the disclosure tag when it is warranted, so the filter should be understood as a meaningful layer of user control rather than a guaranteed verification system.

That limitation is not trivial. Self-reporting only works as well as the reporting. Still, a filter based on declared information can be valuable even without perfect enforcement. It lets a platform expose an explicit preference and gives creators a clearer reason to apply the appropriate disclosure. The alternative—no user-facing control at all—leaves visitors to make guesses from thumbnail art, descriptions and suspicion, which is a poor discovery experience for everyone.

Why this matters especially on Itch.io

Itch.io is a game-hosting platform with an unusually broad range of projects. Its catalog includes the kinds of small, unfinished, playful, strange and highly specific works that do not always fit neatly into conventional storefront expectations. That openness is a strength, but it also means search and browsing tools carry real weight. When there are millions of possible results, curation is not only editorial; it is also the set of controls a visitor can use to shape a feed.

The platform is also widely used for game jams. A game jam is an event in which people make games within a limited timeframe, usually around a prompt, theme or set of rules. Individual jams can take very different positions on generative AI. Some prohibit it, while others may allow it or even center their event around it.

That variation makes a platform-wide discovery preference particularly sensible. A game-jam rule determines what is permitted for entries in that one event. The No AI filter addresses a different question: what an individual player wants to encounter while searching the broader platform. It does not erase the ability of organizers to set their own terms, and it does not force users to opt out of AI-related work. It gives each person a route around it if that is their preference.

A practical change for players looking for specific kinds of work

For someone browsing casually, the feature is straightforward: turn on No AI, then continue using the normal search terms and categories that matter to you. For someone seeking a jam entry, a niche tool, an experimental interactive story or a game by a particular developer, the same filter can be combined with that narrower search.

• Use it as a discovery preference, not a quality seal. A filtered listing is not automatically better, more polished or more ethical in every respect. It only reflects the disclosed-AI preference expressed by the search setting.
• Remember the self-reporting limitation. The filter can only act on the information attached to listings. It cannot by itself guarantee that every result matches every user’s personal standard.
• Read project pages when details matter. A toggle is useful for broad sorting; a game’s own description remains the place to understand what it is and whether it suits your interests.
• Check the rules of individual jams. If a game’s event context matters to you, the relevant jam may have its own policy on generative-AI use.

This approach is less glamorous than a sweeping promise to solve platform spam forever, but it is more immediately actionable. The user gets an option at the point of discovery. The creator gets a reason to disclose. And the platform has a visible interface for an issue that otherwise tends to become a frustrating hunt through individual listings.

Steam remains the obvious comparison

Steam, the considerably larger PC game marketplace, also has rules governing disclosures of generative-AI use. It does not currently offer a store filter that lets users remove those games from results.

That contrast helps clarify why Itch.io’s change is notable. Rules and disclosures are only part of the chain. A disclosure that players cannot conveniently act on has limited value as a discovery tool. A filter translates that information into a decision a person can make in seconds.

There are difficult questions beneath the interface: what counts as relevant AI use, how disclosures are checked, how marketplaces respond to inaccurate labeling, and whether people want a broad exclusion or more granular options. None of those questions is fully answered by a single toggle. But the toggle establishes an important baseline: users should be able to search a game catalog in a way that reflects their own boundaries.

The question is not isolated to games. Platforms in entertainment and technology are increasingly grappling with how AI-mediated recommendations, creation and labeling affect the choices people see. For another example of AI entering discovery experiences, see Netflix’s AI Discover Tab. The difference here is that Itch.io’s new tool is about giving users an option to exclude a disclosed category of work, rather than asking AI to make the recommendation.

Analysis: a filter is a platform design decision, not just a menu item

Itch.io’s implementation recognizes that the debate over generative AI is not only about what creators may use. It is also about what audiences can find, avoid and evaluate. The huge gap between the unfiltered and filtered game totals shows why this becomes a design issue: at that scale, browsing without controls can mean being steered by volume rather than by interest.

There is also a useful neutrality in providing the setting. The platform does not need to declare that every use of generative AI is prohibited in order to acknowledge that many players would rather not browse it. Conversely, people who want access to the full catalog retain it. The decision is moved closer to the user.

Its success will rest on the integrity and clarity of disclosures. Accurate tags make the No AI setting more trustworthy; inconsistent tags make it porous. Yet even with that caveat, the addition changes the default experience for anyone who has felt buried by automatically generated material or simply wants to prioritize projects made without it. For a storefront built around discovery, that is a substantial practical upgrade.