Big Walk AI Disclosure Debate, Steam Reviews and House House’s Response

Big Walk has become the latest flashpoint in gaming’s argument over generative AI, but the available details make this less straightforward than a simple “AI made the game” narrative. House House says large language models, or LLMs, were used only in small parts of the co-op game’s development—amounting to well under 0.1% of the overall code. The disclosure nevertheless prompted a backlash on Steam, a dispute involving Independent Games Festival jurors, and public criticism from Cliff Bleszinski, lead designer of the original Gears of War trilogy.

At the time reflected in the available Steam figures, the game’s overall user rating had recovered to Very Positive, while recent reviews stood at 87% positive. That rebound does not settle the underlying disagreement. It does, however, illustrate that players are weighing at least two separate questions: whether AI should have been used at all, and whether the studio explained that use clearly enough before it became a public controversy.

What House House disclosed

House House publicly acknowledged the use of AI tools after a group of Independent Games Festival jurors said they would not vote for productions that used the technology under any circumstances. The studio’s central claim is narrow: LLM use was confined to minor elements and represents less than one-tenth of one percent of Big Walk’s codebase.

An LLM, short for large language model, is a system that generates or processes text-based responses from patterns in its training. In a development context, that can refer to a broad range of uses, from producing or revising code to troubleshooting an issue through a conversational interface. The supplied details do not establish every task for which the tools were used, who performed the work, or which specific code was affected. They do establish that House House characterizes the use as limited rather than as a means of generating the game wholesale.

That distinction matters to the conversation, even if it does not resolve it. Some players and awards jurors appear to apply a bright-line standard in which any generative AI involvement is unacceptable. Others distinguish between a game substantially created with generative output and a team employing an LLM in a small technical capacity. Both positions can lead to the same request—clear disclosure—but they arrive there from different principles.

The disclosure issue is more complicated than the percentage

The strongest practical criticism is not necessarily the figure House House gave, but when and where players learned it. Big Walk did not initially carry an AI-related label on its Steam page. House House said it had not made a disclosure there because Steam did not request disclosure of code or chatbot use, adding that it was making the statement publicly because the subject had become a matter of public interest.

At the same time, Steam’s rules are widely understood by players as requiring developers to disclose AI involvement. The apparent gap is therefore about definitions and scope: a general expectation of platform disclosure on one side, and the studio’s position that its specific code or chatbot use was not information Steam’s disclosure process requested on the other.

For players, that distinction may feel technical, but it is the heart of the dispute. A disclosure system is only useful when developers and buyers can tell what belongs in it. If a developer believes a tool-assisted code fix does not fall within the requested categories while users assume every form of AI assistance must be listed, a later clarification can look like an omission even where the developer did not view it that way.

This is why the scale of House House’s stated use and the lack of an initial Steam label are not mutually exclusive issues. A person can accept that the use was minor and still prefer that it had been explained earlier. Conversely, a person can dislike generative AI in general but conclude that a tiny technical use deserves a different response than an entire game built from generated art, writing, voices or systems.

Bleszinski calls the response review-bombing

Cliff Bleszinski framed the negative reaction as review-bombing, arguing that people “need to get over themselves” over AI assistance used to get the game finished. In follow-up remarks, he contended that AI is already widespread among developers and that studios are reluctant to acknowledge its use because of the reaction it can provoke.

“Review bombing” is a loaded term. In ordinary usage, it means a concentrated wave of low user scores driven primarily by an external controversy rather than a buyer’s judgement of the game’s quality, performance, design or value. It is not a formal verdict on whether every negative review is illegitimate. Players may sincerely believe a studio’s AI practices are relevant to a purchase decision, while other players may see a flood of low scores over a limited development tool as disproportionate to the game itself.

The term is particularly difficult here because the available evidence shows a real policy and trust debate, not merely an unrelated campaign. The better question is not whether every critical review fits a single label. It is whether a Steam rating is functioning as an evaluation of Big Walk as a co-op game, an enforcement mechanism for an anti-AI consumer stance, or a mixture of both.

The recovery to a Very Positive overall rating and 87% recent positive reviews suggests the latter group was not the only one speaking. Some recent reviewers argued that the situation had been exaggerated. Another response drew a line between using generative AI to construct a game from scratch and using it to address a bug, describing the latter as a more defensible use case. Those comments are individual player views, not verification of the exact technical circumstances, but they show why the review score rebounded rather than moving in only one direction.

Why the IGF juror stance raised the stakes

The jurors’ statement matters because awards conversations often operate by standards different from store policy. A storefront may focus on consumer information and platform compliance. A jury may take an ethical or artistic position that excludes any work involving a certain technology, regardless of how minimal its role was.

That does not mean either framework is automatically wrong; they answer different questions. Store disclosure asks what a purchaser should be told. An awards boycott asks what practices a juror is willing to endorse. The House House disclosure was made in the wake of the jurors’ no-exceptions position, placing Big Walk at the intersection of those two systems.

For independent teams, the consequence is that “we used very little AI” may not be enough to satisfy every audience. A quantitative answer—under 0.1% of the code—addresses scale. It does not answer questions about a tool’s training, labor concerns, creative authorship, or whether there should be a no-use rule in the first place. Those concerns are broader than the supplied information about Big Walk, so they cannot be settled by its code percentage alone.

What a useful disclosure would accomplish

Whatever a player’s stance on AI, the Big Walk discussion points to a practical need for plain language. The most valuable disclosure is not a vague assurance that AI played a minor role. It identifies the category of use, makes clear whether generated material appears in the shipped game, and explains the limits of the tool’s involvement without claiming more certainty than the studio can support.

Category: Was the tool used for code, text, art, audio, customer support, or another purpose?
Scope: Was its contribution limited, as House House says Big Walk’s was, or integral to the production?
Player-facing impact: Did generated output become part of the game players receive?
Timing: Was the information accessible before purchase and before an awards or community dispute made it unavoidable?

Clear answers reduce the chance that consumers must infer too much from a single “AI used” label. They also protect developers from having minor and major applications treated as identical by default. Transparency will not create consensus among people who reject any generative AI use, but it can ensure the disagreement is about the actual practice rather than a vacuum of information.

A narrow case that exposes a broad industry problem

Big Walk’s episode is a useful example of why the industry’s AI arguments keep escalating. One camp sees the disclosed use as limited tool assistance and views the hostile review response as an overreaction. Another sees any use as a meaningful ethical line, particularly when disclosure was not prominent from the start. A third may be less focused on AI itself than on whether Steam’s disclosure expectations were met.

Those positions cannot be collapsed into a single number or one review average. The under-0.1% claim is relevant context, but it does not eliminate players’ right to factor development practices into their buying choices. Equally, a player-led campaign against a game can obscure meaningful distinctions between different kinds of AI use if it treats all involvement as equivalent.

For now, Big Walk’s recovering Steam response indicates that many users are unwilling to judge the co-op game solely through the controversy. The larger lesson for studios is simpler: disclose early, describe the use specifically, and do not assume that a technical distinction will be obvious to an audience that expects direct answers. The debate around AI in games is not disappearing, as broader industry conversations continue to show, including recent discussion around major recognition and expectations within game development. What developers can control is whether players first encounter the facts in a clear statement or in the middle of a backlash.