Steam’s stance on AI-generated content in games has been understandably unclear over the past year, but a new update from Valve has confirmed in no uncertain terms that the company plans to fully support the use of AI in games so long as certain guidelines are followed.
The update came on Jan. 9 when Valve shared a news post on the Steam website to detail its plans to support and monitor games that utilize AI in their development on Steam. Before the update, Steam had not taken a firm stance on whether it supported or condemned AI-generated content, though the removal of a game from the Steam store in June had led to what many were calling a “blanket ban” on AI in games. On June 6, an indie developer explained on Reddit that its game had been removed from Steam and showed receipts of a message from Valve offering the following explanation: “At this time, we are declining to distribute your game since it’s unclear if the underlying AI tech used to create the assets has sufficient rights to the training data.”
While this statement didn’t explicitly say that Steam would not allow AI-generated content, it is the statement that led to assumptions that Steam was placing a blanket ban on AI-generated content, as the truth is that there isn’t any great way to know whether or not an AI has trained on copy-written content or not. With the new policies outlined in Steam’s Jan. 9 update, it seems the burden of proof regarding AI content has been greatly lessened for developers.
The new policy states that Steam plans to release the “vast majority” of games that use AI tools during development. These games will fall into two categories: Pre-Generated and Live-Generated. Pre-Generated games are those that used AI during the development process and include AI-generated art, sound, coding, etc. Live-Generated games are those that will offer real-time AI generation during gameplay and will be capable of offering totally unique AI-generated content for each player. As part of the new Steam Distribution Agreement, developers will “promise Valve that your game will not include illegal or infringing content.”
In June, a game was removed from Steam because there wasn’t any definitive way to prove that its AI didn’t learn from copywritten IPs. Fast forward to January, and it sounds an awful lot like Steam is simply trusting developers to do the right thing when training AI for development. The update continues by saying that games will be examined to ensure these promises are met but doesn’t offer any explanation on how Valve will actually be able to check with certainty that no copyright infringement occurred during AI training.
This allowance of AI-generated content in games has the potential to be a fantastic thing for the gaming industry, as it can potentially allow studios and developers to create their passion projects with significantly fewer monetary requirements. However, if it is not monitored and quality checked, it will simply open floodgates that allow for constant unchecked artistic theft. The truth is that it’s simply too soon to see which direction Steam’s new policy is leaning in, but we’re optimistic that Valve has done its research and has a guaranteed method to make sure art isn’t being stolen by games on its store.
