Steam, a digital distribution platform for video games, holds a vast library of titles that PC players usually flock to in order to procure the latest releases. With the rise of artificial intelligence, though, different developers have come forward to tout the use of such technology in their games, whether by creating backgrounds or writing dialog.
Now, Valve, the owner of the Steam storefront, is purportedly about to bar any games containing artificially-generated content, which could put a spanner in the works for many studios that rely on the distributor.
The news was first shared by an anonymous Redditor claiming to be a developer, who said they had tried to submit a game to Steam but was blocked. Later, they admitted to having used Stable Diffusion in creating their project.
PSA: Valve has been quietly banning newly submitted Steam games using AI-created art assets - if submitters can't prove they have rights for the assets used to train the algorithms: https://t.co/WhWpJpaFjlpic.twitter.com/5KhzJESYxk
In response, a moderator from the platform supposedly noted that the game could not be listed as it used AI-generated art assets derived from third-party sources.
According to the claim, Steam willonly allow AI-generated work if the developer can prove that the data it was trained on was not copyrighted. This seems like an almighty task, especially for more indie companies looking for extra help. The original poster went as far as to inquire whether completely starting over from scratch would be necessary to establish if a work was indeed AI-generated, or if creators could provide evidence some other way.
There could be a workaround for this issue with new tools such as Adobe Firefly, which is trained on public domain and officially licensed images.
According to Forbes, Valve notes that it’s hard to determine where these AI assets have come from, and for now, it’s not accepting any game that uses them. This policy might not affect studios like Blizzard, which uses its data to train its machines.
However, it does seem that new legal cases concerning such systems are surfacing regularly; just last week, two lawyers were fined when ChatGPT provided fake evidence for a case. It might not be the same as generating art or dialog, but it does fall into the same vein of whether AI can be considered fair use. Unfortunately, it will still take time to fully iron out all the kinks.