Valve just changed the rules on AI disclosure for Steam games, and the new policy is... well, it's complicated.
Instead of a blanket "does this game use AI?" checkbox, Steam now has three specific categories developers need to address:
1. General AI content generation - whether AI creates content for the game, store page, or marketing 2. Pre-rendered content - AI-generated assets created before release 3. Live gameplay generation - AI creating content or code during actual gameplay
The big change? Developers now only have to disclose AI publicly if it "ships with your game, and is consumed by players." Translation: AI tools used purely for development don't require disclosure.
So if a studio uses AI to debug code, generate concept art, or speed up asset creation - stuff that never makes it into the final game - they don't have to tell anyone. Only AI that players actually experience needs to be flagged.
This is going to be controversial.
On one hand, it makes sense. Using AI to optimize your build pipeline or catch bugs is fundamentally different from shipping a game where AI generates quests on the fly. The player experience is what matters, right?
On the other hand, where exactly is the line? If a developer uses AI to generate 10,000 texture variations and then manually selects the best 100 for the game, does that count as "consumed by players"? What about AI-assisted writing where a human heavily edits the output?
The policy gives studios a lot of wiggle room, and in an industry that's already sketchy about disclosures, that's concerning.
The timing is interesting too. This comes as AI tools are becoming standard in game development pipelines. Unity, Unreal, and basically every major engine are integrating AI features. Photoshop has AI fill. Audio tools use AI cleanup. It's everywhere.
