The studio's game, 'Clair Obscur: Expedition 33', had its 'Game of the Year' and 'Debut Game' awards rescinded.
However, a prior interview with a producer had acknowledged the studio's experimentation with AI technology.
Unlike traditional 'ugly' placeholder art, AI outputs can appear high-quality.
It's like a wolf in sheep's clothing, blending in where it shouldn't.
This trend increases the likelihood of accidental inclusion of AI assets in final builds.
External vendors may use AI without the client's knowledge. This necessitates that provenance records and contractual disclosures cover all third-party deliveries.
This register would track asset origin, creator, and removal milestones. It should be paired with a strict 'ship gate' requiring manual sign-off for temporary assets.
Forcing placeholder assets to be visually distinct, such as using neon pink textures, ensures they are identified.
Effective Strategy. Visually distinct placeholders are the most effective for identification and removal.
The recent revocation of two Indie Game Awards from Sandfall Interactive’s Clair Obscur: Expedition 33 highlights a critical production failure in modern game development: the inability to track generative AI assets. While the studio maintained that the game was human-made, the awards committee rescinded the honors after discovering that experimental AI-generated textures from 2022 had inadvertently remained in the final build. This incident mirrors similar oversight issues at other studios, such as 11 bit studios and Pearl Abyss, where AI-generated placeholders slipped into released products.
The core issue is that traditional placeholder art was intentionally designed to be visually jarring—often using neon colors or crude shapes—to ensure it was easily identified and removed before launch. Generative AI, however, produces assets that appear polished and finished, allowing them to bypass standard quality assurance checks. Because these assets look like final work, they often evade detection by testers who are focused on functional gameplay rather than asset provenance.
The thesis suggests that the primary failure is not the use of AI, but the lack of rigorous internal record-keeping. When studios submit games for awards or distribution, they often rely on informal memory rather than a centralized, audited register of asset origins. This leads to inaccurate declarations that can result in severe reputational and professional consequences. To mitigate this, studios should implement a formal asset registry that tracks the creation method of every placeholder, paired with strict "ship gates" that mandate the removal of all temporary assets before release. By treating provenance as a core component of production governance, developers can avoid the risks associated with undocumented AI experimentation.