Big Walk AI Disclosure Sparks Debate Over Awards and Transparency

House House’s acclaimed Big Walk has become the center of an uncomfortable industry argument after the studio disclosed that generative AI was used in a limited part of the game’s development. The studio says the use concerned “well under 0.1%” of the game’s code and that generative AI was not used in any other part of production.

The numerical claim will matter to some people. The timing will matter just as much to others. This information was not publicly disclosed at launch or on the game’s Steam page, and it surfaced amid deliberations about the game’s eligibility for the 2027 Independent Games Festival. That puts three separate questions into one knot: what a minimal AI-assisted coding use means, what buyers have a right to know before they buy, and whether awards bodies should treat any generative-AI use as a disqualifier.

None has a painless answer. But treating them as the same question would make the conversation less useful.

What House House says happened

The core of the disclosure is narrow. House House says a fraction well below 0.1% of Big Walk’s code had anything to do with generative AI, with no generative-AI use in other contexts. The discussion around the disclosure characterizes that use as connected to a programming problem, rather than game art, writing, voice work, marketing, or broad production work.

That distinction is significant because “AI in a game” is not a precise description. It can refer to generative systems used to create player-facing images, text, voices, or assets. It can also refer to an LLM, or large language model, used as part of a programmer’s workflow to suggest code, explain an error, or offer a potential fix. Those practices can raise overlapping concerns about labor, training data, energy use, authorship and accountability, but they are not identical production choices.

Nor does a very small code percentage automatically settle the ethical question. A percentage tells us about scope, not whether someone considers the underlying practice acceptable. For players and developers who want a total line against generative AI, the relevant fact is not how much code was involved. It is that the tool was used at all.

For others, context affects what response is proportionate. They may see a one-off, limited coding use as materially different from a project made around generated art or text. That is not an argument that the use should be ignored; it is an argument that a standard for disclosure and a standard for punishment need not be identical.

The disclosure problem is bigger than one percentage

The sharpest reaction is understandable because the information arrived after people had already formed opinions, made purchases, and celebrated the game. House House has said Steam does not request disclosure of code or chatbot usage. Even if a storefront does not require a field for that information, players who want to avoid generative AI can reasonably feel that a meaningful choice was unavailable to them.

Disclosure is not only a warning label. It is the basic mechanism that lets different consumers act on different convictions. A player may decide that limited AI-assisted code does not outweigh the rest of a game’s work. Another may feel that any generative-AI involvement is a boundary they will not cross. Without clear, voluntary information, neither person gets to make a fully informed decision.

That is why a more useful policy would be plain language rather than vague branding. A meaningful disclosure could distinguish among the following:

generative AI used for player-facing material, such as artwork, dialogue, audio or localization;
generative AI used for internal development work, such as code suggestions or troubleshooting;
the part of production affected, stated as clearly and narrowly as possible; and
whether generated output was directly included in the released game.

Those categories would not tell players what they ought to think. They would make it possible for players to decide. They would also reduce the current incentive for studios to hide behind the broad, nearly useless phrase “AI was used.”

There is a practical complication: developers increasingly work in software ecosystems that surface AI functions by default. The discussion around Big Walk points to coding tools, image-editing software and game engines that may embed or promote such capabilities. Avoiding those features can be difficult, especially when the same companies control the tools teams depend upon.

Difficulty is not the same as impossibility, and it does not erase a studio’s responsibility. It does explain why pressure directed only at a small developer may fail to change the conditions that normalize AI-assisted workflows. Game-engine makers, creative-software companies and storefronts all shape what feels routine, what can be disabled, and what creators are encouraged to disclose.

Why the IGF debate is not merely symbolic

The disclosure reportedly became a point of contention among Independent Games Festival jurors considering the game. Some participants favored a firm no-exceptions rule on generative AI. Others argued that the limited use described by House House should be weighed when judging eligibility. The stakes are real: the IGF’s Seumas McNally Grand Prize is its top honor, and Big Walk was viewed as a likely contender.

Awards are not simply lists of games people enjoyed. They are endorsements, and organizations have the right to define the values attached to that endorsement. A festival can decide that any generative-AI use conflicts with its mission, just as it can set rules about release windows, team eligibility or other requirements.

However, a new bright-line rule should also be legible and consistently applied. If an event intends to exclude all generative-AI use, it needs definitions that deal with real development workflows rather than only obvious examples such as generated character art. Does using an AI coding assistant count? Does a feature embedded in commercial creative software count if the developer did not intentionally invoke it? What evidence is requested? How will the policy be enforced across entrants of radically different sizes?

These are not excuses to avoid a standard. They are the work required to make one fair. A policy that exists only after a controversy risks looking arbitrary. A policy published in advance gives studios an opportunity to comply, decline to enter, or ask for clarification before a project becomes an awards flashpoint.

There is also a useful separation to preserve: eligibility is not the same thing as artistic judgment. An awards program could determine that Big Walk is ineligible under a zero-AI policy without claiming that every part of the game is without value. Conversely, a critic or player can continue to find the game artistically meaningful while believing it should not receive an institutional honor. Personal appreciation and formal recognition are related, but they are not interchangeable.

The danger of turning nuance into absolution

Calls for nuance can be abused. Companies routinely use ambiguity, tiny percentages and technical jargon to minimize decisions that deserve scrutiny. “It was only a little” should never become a universal escape hatch, particularly in an environment where developers, artists and players are trying to establish whether generative tools belong in their work at all.

But the opposite shortcut has costs too. Calling every project touched by an LLM identical to a game built from generated assets flattens meaningful differences in practice. It may satisfy the demand for a clean verdict, while doing little to improve disclosure, tool design or labor conditions. It may also concentrate backlash on the rare studio that acknowledges a limited use while leaving larger organizations with far more power—and less transparency—untouched.

That does not make House House blameless. Its disclosure has understandably altered how some people see a game they loved and a studio they had trusted. Trust is partly built by a project’s visible craft, but it is also built by the information a team volunteers when it knows that information could affect a supporter’s choice.

Still, the constructive response is not to pretend the distinction between limited code assistance and wholesale generated production does not exist. It is to insist that both can be disclosed honestly, that both can be debated openly, and that institutions establish their standards before judgment day.

What players, studios and platforms can take from this

The immediate facts around Big Walk will remain a difficult test for anyone who cares about both the game and a stronger anti-AI norm. The longer-term lesson is clearer: secrecy and vague policy create a vacuum, and that vacuum will be filled by suspicion.

For studios, the most credible route is specificity. If generative AI was used, say where, why, and whether its output entered the shipped game. If it was not used, a voluntary statement can be valuable as players look for a signal that is more meaningful than silence. Future commitments should be concrete enough that audiences can evaluate them.

For award organizers, rules should be published before submissions open and should define what is prohibited, what is disclosed, and how edge cases are reviewed. The aim is not to create a bureaucratic trap for independent teams. It is to ensure that a public honor represents a public standard.

For platforms and toolmakers, the challenge is structural. Storefronts can create disclosure options that distinguish code assistance from player-facing generated material. Engine and software providers can make AI features explicit and optional rather than something creators must constantly detect and avoid. Those changes would help developers who want to work without generative AI, while also making claims of transparency easier to verify.

And for players, it is possible to be clear without being careless: ask for disclosure, explain why it affects your support, and direct some attention toward the systems that made opacity normal. The industry conversation is already bigger than a single acclaimed game. It is about whether creative labor gets to set boundaries before automated tools become so embedded that refusing them is treated as unusual.

Big Walk has not produced an easy consensus, and it probably should not. The important question is not whether the game can be reduced to a fraction of its code. It is whether developers, platforms and awards institutions will give people enough information—and meaningful enough policies—to decide what standards they are willing to uphold.

For more on the way AI is being positioned inside another high-stakes industry, see Google Aims Gemini at Healthcare’s Administrative Boss Battle.