
Independent development studio House House announced that it had used generative AI during the final stages of development for its multiplayer adventure game *Big Walk*.
Details Disclosed in Submission Documents, Taking into Account Impact on IGF Awards Judging
This title achieved great success as a communication-focused multiplayer adventure in which players explore a vast open world and aim to solve challenges by utilizing conversation, gestures, and items. It also garnered attention for decisions such as choosing not to release the Xbox version as part of the “BDS movement” regarding the Israeli-Palestinian conflict.
This announcement was made in conjunction with the selection process for the “IGF Awards,” a game award program for indie titles, and the developers stated that they made limited use of large language model (LLM) AI as a means to resolve difficult technical challenges and bugs.The developers state that AI was used in less than 0.1% of the game’s total code and that generative AI was not utilized in any other aspect of the game.
Regarding why they had not disclosed this information previously, the company noted that Steam had not required them to disclose details about their code or chatbot usage. However, after a joint statement by some judges revealed that the use of AI significantly influenced the IGF selection process, they decided to make the information public as a “matter of public interest” by disclosing the details included in their IGF submission.
Apology in Support of Efforts to Eliminate “AI Harm”—Explicitly States That Developing Without a Clear Usage Policy Was a Mistake

House House stated that its position is to “support the efforts of those working to limit the harm that generative AI causes to video games and the world,” and apologized for the possibility that this announcement fell short of the standards set by those efforts.Expressing its view on the matter, the company stated that it does not endorse or advocate for the use of AI tools, even in minimal ways, and expressed concern that “voluntary disclosure might lead to this technology becoming normalized or treated as ‘inevitable.’”
They also explicitly stated that developing *Big Walk* without first establishing an internal AI usage policy was “a failure.” They added that they will strive to formulate a more comprehensive policy moving forward.

It was noted that the sections actually modified using an LLM were code intended to resolve issues—such as “the game running 50% slower for about 20 seconds at sunrise and sunset”—that originated in third-party code. After consulting with the third-party engineers, it was concluded that these issues could not be resolved, and the LLM-generated code was used to address them.Technical details were also revealed, including that the code worked successfully after significant modifications were made to the solution proposed by the LLM.







