Support us on Patreon
PayPalWith more and more game developers creating AI slop, or just using AI generation for stupid things in their games, we need more ways to avoid them.
The problem is for me, I'm so busy that it's just easy to miss. The AI notice on Steam page is buried is small writing at the bottom of a store page. Steam itself has no filters for it currently either, although the community site SteamDB does and the list is a lot bigger than I expected. 12,284 games on Steam have an AI disclosure according to SteamDB at time of writing. I was going to say "that's a crazy number!", but it's not really is it? I'm sure many GamingOnLinux readers fully expected a lot of developers to end up pushing out slop with AI generation.
Thankfully, an open source browser userscript was pointed out to me on Bluesky (thanks!) which really makes it impossible to miss. It will work on any operating system with pretty much all major browsers. With that installed, when going to a Steam page that has an AI disclosure, you get this appear before you're able to actually do anything:
I love how simple and most importantly how effective it is. The main thing is that doesn't stop you taking a look, it just makes sure you're fully aware of the AI gen use. Hopefully never again will I completely waste my time playing through a game, writing it up and then suddenly remembering - "Oh yeah, I have to check if they stuck AI generation in it".
I'm sure some readers might find my stance on AI generation a bit harsh perhaps, but when there's tens of thousands of games releasing every year — my time (and yours!) is a premium currency. Why would I want to spend my time with a game full of AI generation, when I could play one that had real actual people craft the work. I can just skip over a game with it for a different game without it. Some of the AI use just seems completely daft as well, one I saw earlier noted how it was used for "Icons + Artists Paintover for 5 icons, out of 100s of artist icons" (Kingdoms Reborn) and in that case I think like…why was it even needed at all?
Using this browser userscript, and remembering to check SteamDB, it's a genuine surprise to see just how many titles have now added an AI disclosure. To my surprise even games like My Summer Car, The Outlast Trials and Siralim Ultimate have the notice.
Hopefully you find it helpful. With tools like the recent EdenSpark announcement, we're going to see more AI games.
Article taken from GamingOnLinux.com.
See more from me
The trouble is, they're conflicted. On the one side is copyright and their ability to squeeze all the money out of intellectual property forever.
On the other, they might get to use the technology to replace all the artists, writers, actors, etc. with free robot slaves. Or worse, someone else might get robot slaves first.
It's a tough call.
ah, the old shall i be evil or even more evil thought process.
Last edited by Lofty on 22 Oct 2025 at 12:26 am UTC
TheRiddick 15 hours ago
I guess if it was used for texture creation it can also be quite good.
People need to remember that games have been doing procedural generation for a long time before this AI craze and it is very similar sort of thing.
Wish we could identify AI slop more accurately rather then just labelling everything that uses AI in some way somewhere being bad, because it won't be long before all games made use it in some form!
Last edited by TheRiddick on 21 Oct 2025 at 10:42 pm UTC
I'm not opposed of AI gen being used for voice acting (when done right and not stealing VA)
define stealing. Outright plagiarism is perhaps easy to spot but hard to prove let alone raise an expensive legal challenge.But taking a VA and modifying it is even more deceptive. The voice isn't just an audio signature , it's the soul & passion of the artist.
back to first point .. this is how i see what's happening.
They know there stealing
We know there stealing
They know we know there stealing
But what is a poorly paid struggling artist going to do against an army of billionaire backed lawyers ?
as @Nezchan put it there probably vacillating between litigating for billions themselves or working out how many billions they can save by wiping out large portions of 'useless' humanity using Ai
Last edited by Lofty on 22 Oct 2025 at 7:54 am UTC
In The Finals they use it for the voice of the match commentators with mixed results... And most of the time you are too focused or talking to your team mates to pay attention to them.
In The Finals they use it for the voice of the match commentators with mixed results... And most of the time you are too focused or talking to your team mates to pay attention to them.
but who's voice signature was it trained from ? would it be possible to ask the studio ? if there is no reply could you use a FOIA request to find out ? what's the law around this , does anybody know ?
What i don't understand is where the lawyers at ?
I believe the problem is that buying a piece of media and feeding it into an algorithm to create a model isn't considered copyright infringement because the result is a piece of software rather than a copy of the original piece of media. Yes, if you ask a picture generating model to generate a picture of an archeologist with a hat who uses a bullwhip it generates an image which is easily infringing on a well-known intellectual property, but then it's still not the model which is infringing. Now, machine generation of media is a tool, so creating an infringing text or tune or image isn't a problem per se. If I put a bunch of cans of paint on a shelf and then knock the shelf down and the paint splatters form an image of a familiar Italian plumber with a red hat I haven't done anything wrong. Distributing the result, however, will likely get me into hot water (if the rights holder decides to bother itself with me).
Typing this comment got me thinking, though, this kinda reminds me of software libraries. Machine generating models are basically turning various forms of media into something like software libraries from which pieces can be reused to make something new (in as far as anything can be new when, of course, all original ideas have long been had and we're just rehashing the same stuff over and over again). Back in the day software didn't have licenses and you could freely reuse things from programs made by others but those days came to an end. Maybe we'll see media start getting licensed to not allow machine generation model creation.
I believe the problem is that buying a piece of media and feeding it into an algorithm to create a model isn't considered copyright infringement because the result is a piece of software rather than a copy of the original piece of media.You're probably right, but that just means that copyright law needs to be updated.
If they want to create a new AI that's a great interior designer, but the creators of that AI need to knock doors and be shown the insides of people's houses - let's face it, they'd be told to do one.
The issue here is that they didn't have to knock doors and be let in. They just scraped everything they could find on the internet and considering all that data "free".
Even when it was provably not free, such as when Meta, absolutely scumbags that they are, scraped terrabytes of LibGen for pirated books, to train their shitty model. And then the ball-less judge ruled against the authors. Nothing about them breaking the law by torrenting LibGen, nah, just a weak excuse about "not dilating the market through their actions". Holy fuck, Zuckerberg makes me feel physically sick.
If you're still using Facebook or Insta after this, you have no soul. I can't judge you for Whatsapp because I know from personal experience how difficult it is to get people moved to alternatives, when they just don't care.
https://www.theguardian.com/technology/2025/jun/26/meta-wins-ai-copyright-lawsuit-as-us-judge-rules-against-authors [External Link]
Remember Kazaa in the early 2000's? Remember that single mother being sued into oblivion because her son shared songs using it??
In Duluth, Minnesota, the recording industry sued Jammie Thomas-Rasset, a 30-year-old single mother. On 5 October 2007, Thomas was ordered to pay the six record companies (Sony BMG, Arista Records LLC, Interscope Records, UMG Recordings Inc., Capitol Records Inc. and Warner Bros. Records Inc.) $9,250 for each of the 24 songs they had focused on in this case.
(Where are those guys? She ended up owing over £2M - which was reduced to around £45K after YEARS of retrials because who the fuck thinks that anyone should be fined £2M for downloading/sharing files in 2003. Insane.)
Anyway. Clearly anything on the internet is free now. So that's nice.
Last edited by scaine on 22 Oct 2025 at 1:35 pm UTC
How to setup OpenMW for modern Morrowind on Linux / SteamOS and Steam Deck
How to install Hollow Knight: Silksong mods on Linux, SteamOS and Steam Deck