Steam has introduced a new policy requiring developers to disclose the use of any artificial intelligence (AI) generated art, music, or code in their games. This policy applies to both pre-generated content and materials created in real-time during gameplay. Valve – Steam's owner – has been evolving its policies on AI since last year.
Last July, the company clarified its position as it stood at that time. Valve clarified its stance on AI art on Steam, following concerns that the company is banning games that use AI art. The company said there would be no outright ban on AI content but it would be looking at measures to police users. Now it seems Valve has settled upon a fixed policy.
Game Submission Requirements
As per the latest update, developers seeking to launch their games on the Steam platform must now complete an AI disclosure form. This form necessitates developers to ensure that any pre-generated AI materials are free of illegal or copyright-infringing content. Furthermore, games that feature live-generated content must provide a detailed account of the safeguards implemented to prevent the AI from producing inappropriate or illegal content on the fly.
Valve has also encouraged its user base to play an active role by reporting any illicit AI-generated material they encounter within games. This reporting mechanism is intended to maintain vigilance among developers and ensure compliance with the new policy. In light of these recent changes, Valve has reached out to game developers who are active in the AI domain and those spearheading AI tools' development, iterating that the goal is to foster a more open approach to AI-centric games on their platform. However, it was explicitly mentioned that Adult Only Sexual Content created with live-generated AI is not permitted for release at this time.
AI and Legal Scrutiny
This regulatory move follows heightened concern across the industry about potential legal ramifications associated with hosting synthetic media that may breach copyright laws. Leading AI model developers, including OpenAI, Google, Microsoft, Midjourney, and Stability AI, have been embroiled in legal disputes over the alleged use of copyrighted material in the training and operations of their neural networks — claims these companies deny.
In response to worries about the legal liability of AI-generated content, tech giants such as Microsoft and Google have pledged to defend their users against any copyright infringement legal actions spawned by the output of these AI models, though specific conditions apply. The question of whether the utilization of copyrighted content to train such models falls under the banner of fair use remains unresolved in legal spheres.
Valve has acknowledged the complexity inherent in these decisions and has committed to ongoing evaluation and learning from both the submissions they receive and the evolving legal context surrounding AI. They have indicated that revisions to the current decision may occur as necessary.