Want to see Raider King content first? Add us as a preferred source.
Only 8% of Steam’s top new sellers in 2025 openly disclosed generative AI use on their store pages, and that contrast, between what’s disclosed and what players suspect is happening behind the curtain, might be the most important part of this whole conversation. Steam’s year-end “Best of” lists are explicitly revenue-based and grouped into tier buckets (Platinum/Gold/Silver/Bronze), so this isn’t some niche sample of tiny releases. These are the real money-makers.
What the 8% actually represents
The heart of the story, for me, is that “8% disclosed” is a clean, concrete number, but it’s also a very specific kind of truth: it tells us how many major 2025 top sellers were willing to put “yes, we used GenAI” on the label where customers can see it. Valve has a disclosure system in place and has publicly talked about requiring developers to disclose AI usage, including distinguishing between pre-generated AI content made during development and live-generated AI content created while a game is running. That means the label is supposed to exist for a reason, and it’s supposed to help players make informed choices.
In your disclosed bucket, the names are the attention-grabbers people will recognize: EA Sports FC 26, ARC Raiders, Call of Duty Black Ops 7, inZOI, Anno 117: Pax Romana, Where Winds Meet, Revenge on Gold Diggers, and Jump Space. Those are the games that, at least on Steam’s store-page level, aren’t trying to pretend the topic doesn’t exist. And even if someone doesn’t care about AI discourse at all, that list alone shows the tech isn’t confined to shovelware or “that one weird demo,” it’s sitting right in the same year-end revenue neighborhood as big mainstream releases.
The part that gets messy fast
Where things inevitably get heated is the second layer: the games that didn’t disclose on Steam (or no longer do), but still had credible “AI touched this project” reporting attached to them.
This is where Clair Obscur: Expedition 33, Jurassic World Evolution 3, and The Alters come up, because the way they’re discussed tends to blur together three very different scenarios: experimenting with AI and shipping some of it briefly before replacing it, using placeholder assets and later removing them, or disclosing and then pulling the disclosure after removing the AI-made pieces. That distinction matters, because “GenAI was part of the dev process at one point” is not the same moral claim as “the final shipped game is full of AI-made assets,” even if both will get flattened into the same argument online.
Then there’s the wider “known use” cluster from the same discussion: Battlefield 6, Avowed, Grounded 2, The Outer Worlds 2, Inazuma Eleven: Victory Road, and Escape From Tarkov. Some of these are about visible content controversies (where players point to specific images or assets), and some are more about dev-side workflow stories, the kind of stuff that’s hard for the average player to see unless a developer says it out loud.
Why I think AI use is inevitable
This is the part where I’m just not able to buy the fantasy that game development is going to stay AI-free in any meaningful sense. Even setting aside the obvious stuff like upscaling, cleanup tools, and automation that’s been creeping in for years, the modern production reality is that teams will adopt anything that saves time, reduces repetition, or helps them iterate faster, especially when schedules are brutal and budgets are always under pressure.
Valve’s own stance is basically an admission of that reality: the platform isn’t saying “no AI,” it’s saying “disclose it (if you want to), and be responsible about what you ship.” And outside the top sellers bubble, reporting has suggested GenAI disclosures across all Steam releases have become fairly common, which makes it even harder to argue this is a temporary fad that will just vanish.
So yes, I think some amount of AI usage in game development is going to become normal. The question that actually matters is whether we normalize it in a way that’s ethical, transparent, and respectful to human creators, or whether we normalize it in the ugliest possible way and then act shocked when players stop trusting studios.
The real moral line is how it’s used
For me, the moral argument isn’t “AI is always evil” or “AI is always fine.” It’s whether the use is exploitative, deceptive, or disrespectful to the people whose work makes games worth caring about in the first place.
If a studio is using AI like a rough sketch layer that gets replaced by real concept art, or as a temporary placeholder that never survives to the final build, that’s a fundamentally different moral situation than shipping AI-generated art or voices without disclosure, or training models on creators’ work without permission and then using the results to avoid paying those same creators. And transparency is a huge piece of it: players can’t consent to what they can’t see, and they definitely can’t evaluate what they can’t understand.
Even if someone believes the “true” number is higher than 8% (and honestly it probably is), the disclosure number is still the part with the clearest signal. It shows how many of the biggest releases are willing to say it plainly on the store page, in the same space where they disclose controller support, multiplayer features, and accessibility notes.
And if AI really is going to become a standard part of the gaming experience, then disclosure can’t be treated like an optional PR move that only some teams bother with. It needs to become boring, consistent, and specific, because that’s how you get players to argue about the actual ethics instead of spiraling into guesswork about who’s hiding what.
