This controversy began with an X post about Epic Games CEO Tim Sweeney. He criticized Valve’s requirement that games disclose AI usage on Steam, arguing that this kind of label can give developers a negative mark and affect game launches and sales.
Original post:
|
|
The source of the issue is not just a simple social-media complaint, but a PC Gamer interview with Tim Sweeney. In the interview, Sweeney discussed Unreal Engine 6, game industry costs, AI tools, and Steam’s AI disclosure policy. His core point is that AI has already become part of the development toolchain, and if a platform displays “uses AI” as a prominent label, developers may be judged by players before their work is evaluated on its own merits.
What Steam’s AI Disclosure Is
Since 2024, Valve has required developers to explain in the Steamworks content survey whether a game uses generative AI. Developers need to distinguish two cases:
- Pre-generated content: assets generated with AI during development and eventually included in the game or store page, such as art, text, audio, code, and related materials.
- Live-generated content: content generated by AI during gameplay for the player experience, such as dynamic text, images, dialogue, or other interactive content.
This information may appear in the AI disclosure area of the Steam store page, so players can know before purchase whether a game includes AI-generated content.
Valve’s logic is straightforward: players have a right to know, and the platform needs developers to explain how they avoid illegal, infringing, or harmful content. For live-generated content, Valve also allows players to report potentially illegal AI outputs inside the game.
Why Sweeney Opposes It
Sweeney’s objection is not that AI should never be regulated. His objection is that Steam turns AI usage into a public label.
His concern has three layers.
First, Steam is extremely important for PC game distribution. Many developers cannot realistically bypass Steam even if they dislike some rules. If they want wishlists, visibility, and user reviews, they have to face Steam’s disclosure requirements.
Second, AI labels can become moral judgments. Many players who see “uses AI” do not distinguish whether AI was used for placeholder concept art, code assistance, bulk asset cleanup, or final art replacing artists. A single label can mix very different levels and types of use.
Third, small teams rely more on tools. Sweeney argues that game development costs keep rising, and small teams need productivity gains to compete with long-running large games. If players and platforms stigmatize AI tools as a whole, small teams may struggle even more.
Why Players Are Not Convinced
Player resistance to AI games is not imaginary.
Many players are not simply angry that developers use tools. They worry about these issues:
- Whether AI training data infringes creators’ rights.
- Whether AI assets lower game quality.
- Whether developers use AI to cut costs while still charging premium prices.
- Whether AI-generated content makes games feel cheap, repetitive, and lacking human taste.
- Whether platforms will be flooded with low-cost AI content, making normal games harder to discover.
This is why Steam’s disclosure policy has support from some players. To them, AI disclosure is not punishment for developers. It is basic pre-purchase information, similar to in-app purchases, adult content, violence, or online features.
AI Labels May Really Affect Sales
The Steam AI disclosure debate is intense because it may have real business impact.
Windows Central, citing Game Oracle’s research, reported that researchers analyzed nearly 10,000 Steam games released in 2025 and found that games disclosing AI usage performed worse in first-month reviews, ratings, and comparable conditions. The research also said that, after controlling for developer experience, publisher support, genre, and release timing, games with AI disclosure may receive significantly fewer reviews than similar non-AI games.
This should not be simplified into “players are definitely boycotting AI.” AI usage may also correlate with other problems, such as low budgets, low quality, weak marketing, inconsistent asset style, or developers using AI to quickly pile up content. In other words, the AI label may trigger bias, but it may also reveal quality issues that players can already feel.
The Real Issue In This Debate
The core question is not whether AI can be used in game development. The real debate is about three more practical issues.
First, how specific should disclosure be? Writing only “this game uses AI” is too broad. It can lump together GitHub Copilot, Photoshop assistance, concept exploration, final art, and live dialogue generation. More granular disclosure could reduce unfair harm.
Second, should platforms protect players’ ability to filter? Some players do not want to buy games that include AI-generated assets. Others only care whether the game is fun. If platforms remove disclosure entirely, the first group loses useful information. If disclosure is too broad, labels become stigmatizing.
Third, should developers explain the boundaries of AI use? In the current public-opinion environment, saying “we used AI to improve efficiency” is rarely enough. A better approach is to explain where AI was used, what was completed by humans, whether licensed data was used, whether jobs were replaced, and whether AI output entered the final product.
Advice For Developers
If a game does use generative AI, developers should not treat Steam disclosure as just a compliance form.
A safer approach is:
- Clearly separate internal tools, temporary assets, and final shipped content.
- If AI-generated content appears in the final game, explain where it appears.
- Avoid vague wording such as “AI-assisted development,” which can easily be misunderstood.
- Keep records of asset sources, licensing, human review, and replacements.
- Use store pages, announcements, or FAQs to explain the boundaries in language players can understand.
Players are not naturally against every tool. What triggers controversy is when developers try to hide AI usage or package visibly rough AI content as normal creative work.
My View
Sweeney is pointing to a real pressure developers face: games are getting more expensive, competition is getting harsher, and AI tools can reduce a lot of repetitive work. But Valve’s disclosure direction also makes sense. When players are already sensitive to copyright, job replacement, and low-quality AI content, removing disclosure entirely would only deepen distrust.
The better direction is not to remove AI disclosure, but to make it more specific and easier to interpret. Platforms should distinguish between development productivity tools and AI content that players actually consume. Developers should also proactively explain their AI boundaries. Otherwise, AI labels will remain an entry point for arguments instead of a useful way to understand how a game was made.