The debate over AI disclosure labels in gaming just got a little clearer, at least on Steam. Valve has quietly updated its AI disclosure requirements for developers, specifying exactly when generative AI usage needs to be flagged on the platform.
This update comes on the heels of Epic Games CEO Tim Sweeney’s controversial statement that digital storefronts should abandon AI disclosure labels altogether. His reasoning?
“AI will be involved in nearly all future production” anyway, making such labels pointless. PC Gamer’s Tyler Wilde pushed back on that argument, but he also highlighted a genuine gray area: where exactly do developers draw the line?
Should a game be labeled if someone used Photoshop’s generative fill tool while sketching concept art that never saw the light of day?
A Developer’s Guide to Evolving AI Disclosure Requirements of Steam
What about developers who lean on Claude or ChatGPT to generate a few code snippets? Or marketing departments utilizing AI to manage spreadsheets? Such questions are not merely hypothetical; they are real concerns faced by developers who have to work in an ever more AI-integrated workflow.
The response by Valve, as seen in a screenshot shared by Simon Carless’ newsletter, GameDiscoverCo, brings much-needed clarity on the matter. According to a new form that has been submitted, pre-built assets of a generative AI only need to be disclosed in specific instances: if used for marketing or if used in something that “ships with your game, and is consumed by players.”
This is an important distinction to make: Valve isn’t looking to police every little AI tool that developers might use behind the scenes. Using AI-powered coding assistants to debug your game or generative tools to speed up internal workflows? That’s not what Steam’s disclosure system is looking for. The focus is directly on player-facing content-the final art, sound, music, and writing that actually makes it into the game or its promotional materials.
The disclosure form maintains Valve’s existing two-category framework for AI usage, introduced by the company for the first time in 2024. First, there’s the “Pre-generated” category, which covers any in-game content created with the help of AI tools during development.
Such assets are subject to the same content rules as traditionally created content. The second category, “live-generated,” applies to games creating content through AI systems while the game is actually running think procedurally generated dialogue or dynamically created visuals.
Steam’s Evolving Stance on Generative AI
In the case of live-generation AI game titles, Valve asks for a description of what measures have been put in place by the developer to ensure that the AI in question does not produce illegal content. This is not an unreasonable request, given the propensity of generative AIs to produce troublesome outputs when left unchecked.
In a gaming-related issue with problematic output, Valve has also introduced a reporting function in overlay of Steam for illegal content created using AI. In light of several eye-catching outputs we’ve noticed AI models have exhibited as of late, it seems a reporting function is a prudent move.
The timing couldn’t be more apt. The industry’s now wrestling with how-or even if-it should be transparent about integrating AI within game development. Some players want to financially support titles created fully by human beings, while other gamers simply don’t care about the tool used but about the end product.
Meanwhile, developers get pulled from every direction-investors pushing to see gains in efficiency via AI use and communities that might look upon usage of AI with suspicion or even outright hostility.
How Disclosure Policies are Shaping the Future of Steam
Just recently, one RPG developer had to defend publicly their game from accusations of using AI, saying they “poured years of our lives into this game and only worked with real human artists on everything.” These types of accusations-whether founded or not-show why clear disclosure policies matter.
Valve’s new policy walks this tight line rather well. The company understands that AI tools have become an ordinary part of any game development process without requiring devs to report each individual interaction with an AI system, for instance. The new rule provides gamers with the transparency that they have been demanding.
Whether this solution satisfies all parties is yet to be determined. There is some logic to Sweeney’s point that the labels AI makes no sense, given the inevitability of AI everywhere, but in disregarding the needs of consumers to know what they’re purchasing, he also overlooks the potential benefit to consumers. Valve’s middle-of-the-road solution, in the meantime, appears to be the best course.
What will the future hold as the world of games continues to change under the influence of AI? One thing is sure these polices will continue to adapt and change with the times. Meanwhile, at least Steam developers will know under which circumstances they must check the box about their use of AI.
