
A yellow card from Steam for AI in computer games
One of the significant uncertainties in this process is the question of copyright in game elements created this way.
Already over 94% of IT specialists in Poland use such tools 1 .
One of the significant uncertainties in this process is the question of copyright in game elements created this way.
The creators of the data on which these systems are trained and which they use have often not given their consent, which has opened the door to the first disputes in the USA.
Recently, the community gathered around Valve, the owner of the online game platform Steam, began a discussion about whether Valve does not, in fact, accept games using elements generated by artificial intelligence.
This topic was started by one of the developers on Reddit and quickly gained popularity.
This developer described a situation in which his game could not be approved by Valve because it had “ a few assets that were fairly obviously generated by artificial intelligence ”.
The developer community took this as a signal that Valve may no longer want to publish games with any content generated by artificial intelligence.
Speculation among internet users led Valve to publish an official statement on June 29, 2023 in VGC.
In it, we read that the platform's goal is not to discourage developers from using artificial intelligence, but rather concerns about copyright infringement related to the problematic ownership of elements generated by artificial intelligence.
In Valve's assessment, publishing a game would be possible if its developer could confirm that they hold the intellectual property rights to the dataset on which they trained the artificial intelligence to create assets in their game, quote: “ although developers may use these AI technologies in their work, they may not infringe existing copyrights ”.
The developer's possession of proof of ownership of the data on which the algorithm they use was trained is the ticket to releasing the game.
Such a policy constitutes a significant restriction on the use of artificial intelligence and machine learning algorithms in the game development process.
When using certain tools, it will be very difficult (if at all possible) to ensure that a party holds the rights to all input data.
Requiring developers to submit a declaration that they hold the rights to all the intellectual property used in the game development process is an attempt to shift responsibility for potential copyright infringement from the platforms to the developers.
Presumably, until these issues are clearly resolved in new regulations or in the course of proceedings that are already underway, platforms will make the publication of games conditional on the submission of such declarations.
From a practical standpoint, if irregularities come to light, the platform owner will be able to (reasonably) claim that the developer misled them.
Valve claims that its new policy is dictated solely by the law and not by a general attitude toward artificial intelligence and machine learning.
Moreover, Valve declares that, as regulations change, its internal product verification policies will also be adjusted.
As Valve writes: “ We are still learning about artificial intelligence, the ways it is used in game development, and how to account for it in our process of reviewing games submitted for distribution on Steam.
Our priority, as always, is to try to release as many submitted titles as possible ”.
In the context of Valve's actions, we encourage all developers to verify the extent to which solutions and tools based on artificial intelligence can be used, and to prepare internal manuals and guidelines for their creative teams. https://nofluffjobs.com/insights/ai-na-rynku-pracy/ Have a question?
Let's talk. A 20-minute conversation.
No briefs, no forms.
We will answer directly.
Book a call → See more articles

dotlaw is an AI-native law firm for technology companies. Specializations: AI Act, GDPR, MiCA, ISO 27001, IT contracts, M&A in tech.
Meet the author

