Epic’s Tim Sweeney challenges AI labeling in games as Valve’s Steam enforces disclosures. Explores nuances, industry backlash, and why transparency divides.
© A. Krivonosov
Epic Games CEO Tim Sweeney has reignited debate over how AI should be used and labeled in the games industry. His recent claim that it makes little sense for stores to tag projects as made with AI drew a wave of criticism, yet the remark turned out more nuanced than it first seemed.
The spark was a growing trend of platforms stepping away from the made with AI label. Sweeney argued that such tags are pointless because AI will be involved in almost every future game production, and he hinted at Valve’s practice of asking developers to disclose their use of AI on Steam.
The issue flared after the Arc Raiders and The Finals cases, where teams used generative AI to produce voice-over based on actors’ voices. Players pushed back, but the boundaries of AI use proved fuzzy. In Arc Raiders, for instance, the animation team used AI tools mainly to smooth transitions — more as a helper than a content creator. That nuance is exactly where the conversation tends to overheat.
Valve has taken a more transparent route. Since January 2024, developers must state whether AI is being used and distinguish between content generated ahead of time and content generated in real time. Part of this disclosure appears directly on a game’s store page so players can judge for themselves. According to a Steam survey from July 2025, roughly 7% of projects had already flagged the use of generative AI.
Sweeney’s line spawned multiple interpretations. Online, many read it as a call to ditch labeling altogether and to ignore how AI is used — whether for voice, art, or basic technical assistance. The core problem is that the term AI has ballooned to cover everything from image and voice generation to coding aids and everyday audio simulators that musicians rely on for soundtracks. The line between assistance and authorship is getting harder to draw, and that ambiguity fuels concern more than any single tag can resolve.
Pushback came quickly. Former Counter-Strike artist Ayi Sanchez likened the lack of labeling to an attempt to hide a product’s ingredients. Composer Joris de Man pointed to the precedent of mandatory not actual gameplay notices in trailers. And indie developer Mike Bithell said that if Sweeney believes in AI, he should wear the label proudly and be ready to see sales drop. The subtext was clear: transparency is part of trust.
There were defenders, too. Matt Workman, author of the post that sparked the debate, argued that Steam’s policy is overly broad: by its logic, virtually any developer using Unreal Engine, Google Suite, Slack, Adobe, or Microsoft Office falls under the rule, since all of those products come with AI tools baked in by default. It’s a fair point — when everything contains AI, a blanket label risks saying very little.
The argument over AI and transparency in game development is nowhere near settled. Sweeney’s comment merely underlined how hard it is to draw a clean boundary — and how contested the very idea of a game made with AI has become. What the industry lacks is a shared language for these distinctions, and until that emerges, every label will feel either too blunt or too late.