Steam has a small section that indicates whether a video game uses AI to generate certain types of content. Of course, this largely depends on the honesty of the developer, who may try to hide such information. However, this system had another major problem. The question is: exactly what AI-generated content should be reported?
It looks like Valve has finally answered that question. Only content that players actively experience should be published.
Steam’s new rules include:
In other words, Steam’s AI content disclosure now excludes “behind the scenes” tools used during development, such as AI-assisted coding and production optimization systems. Instead, developers should only flag content that directly impacts the player experience, such as AI-generated graphics, music, sound effects, and text. This includes both in-game content and marketing materials that players see, hear, and read.
The new rules apply to all AI-generated elements, whether created in advance before release or in real-time during gameplay. For example, consider an AI-powered NPC that dynamically responds to player interaction, as in the game Where Winds Meet.
For real-time generated content, developers must also explain what safety systems are in place to prevent AI from creating illegal or inappropriate material. A report button has been added within the Steam overlay, allowing players to flag games that use real-time generated AI to create illegal content.
