The rapid advancement of generative artificial intelligence has outpaced the legal and bureaucratic systems designed to regulate it, creating a complex landscape for the video game industry. While initial industry excitement focused on AI's potential to democratize game development and enable smaller studios to create larger titles, concerns over copyright infringement and ownership are now prompting widespread adoption of "no-AI" clauses in contracts. According to videogame lawyer Haley MacLean, these clauses have become standard practice, particularly over the past year, as companies seek to mitigate legal risks.
MacLean, an IP lawyer and video game specialist at Voyer Law, told GamesRadar+ that while such clauses were rare a few years ago, even smaller studios now routinely include them to "cover their butts." She emphasizes that their prevalence has reached a point where their absence is notable. "It's just boilerplate, to the point that if I did encounter one that didn't have a gen AI clause, if the publisher was creating assets too, I would say we need to push for that to make sure the publisher is not using gen AI in the marketing, porting, QA of your game," MacLean explained.
This shift is influenced by several factors, including the potential negative impact on player reception and the evolving legal understanding of AI-generated content. Data suggests that AI disclosures can significantly reduce the number of game reviews and skew them towards negativity, fostering a community backlash against AI usage. "I don't want to use gen AI because my community would hate it," MacLean noted, reflecting a sentiment that influences developer decisions.
The legal challenges extend to the very concept of ownership. If AI generates an asset, proving that it did not infringe on existing copyrighted works is difficult. More critically, it can be equally challenging to legally assert that the user of the AI tool actually created the asset. MacLean illustrated this with an example: if a game uses AI-generated trees, and these assets are not significantly transformed by human artists, other developers might legally use those same AI-generated trees without repercussions. "Legally speaking, you can't really argue YOU made it, either," she stated.
This creates a paradox for companies aiming to cut costs by reducing artist headcount through AI. To maintain legal ownership of AI-generated assets, remaining artists must substantially modify them. However, a reduced workforce and potential resentment from employees could increase the likelihood of legal disputes or alienate consumers, potentially negating cost savings. "In order to fulfil the promise of generative AI—that you could lay off a bunch of artists and just have your hallucination machine dream something up—you need to lower your headcount. But your remaining artists then need to meaningfully transform what AI generates, in order to legally lay claim to it," MacLean observed.
MacLean anticipates a rise in lawsuits related to these issues as companies continue to seek faster, cheaper development methods. Current legal precedents in the US and Canada hold that AI cannot own intellectual property because it is not human-made. "We're in that frustrating middle with new tech where it's very fun, it's very flashy, everyone's excited, but I think the legal issues are going to just be popping up more and more," she concluded.