It’s also not particularly helpful, right? Because if I use an LLM to debug a crash report, then that’s the same category as creating a slop game.

So what you would do, realistically, is use the AI to write code, debug, you know typical software development stuff and then just… lie. Your game is closed-source anyway, so nobody is reading the code.

There’s levels to this. AI generated slop assets is one thing. AI-assisted coding describes every software project everywhere in the world.

Steam has clear guidelines on what requires an AI disclosure, and all that people are asking for is that they can filter based on this. It's a tiny change.

> AI-assisted coding describes every software project everywhere in the world

That is simply not true.

> and then just… lie

I'm not a lawyer, but lying about something that influences the customer's purchase decision sounds a bit like fraud and not something I would advise to "just" do. If using AI is so normal then why lie about it?

No, it is. You can ask any software developer working at any company right now if they use AI in their workflow somewhere in the company, either them or other developers. They will say "yes". The only people who think LLMs aren't typically used in software are not software developers. Which I get it, if you don't know then you don't know. So ask. It's not just one or two companies, it's also not 90%. It's 99.99%, I would say 100% but I can't say for sure.

> not something I would advise to "just" do.

Not something I would advise either, but people are going to do it. The difference is indistinguishable and you can't verify it anyway because you typically can't access the source code. So, of course people will lie.