Generative AI in game development: Navigating IP, regulation, and reputation

Ali Mirsaidi is a partner at the international law firm Perkins Coie and a member of the firm’s Digital Media, Entertainment, and Gaming practice group.

This article is part of AI Week.

Over the past few years, AI use in gaming has taken on new meaning and new controversy. A constant stream of headlines reflects heated debates about what role generative AI will play in game development, with community members’ discomfort rooted in a perception that these tools were trained on artists’ work without consent – a concern that can quickly translate into reputational fallout for studios that deploy AI systems. Community backlash can be swift, shaping audience trust and commercial outcomes.

While some of the ongoing discourse creates the impression that studios are merely experimenting with these tools, the reality is that AI is already part of the core infrastructure for the industry, subtly embedded in workflows across game production. In 2025, one in five games released on Steam disclosed generative AI usage, and that number will only rise in the years to come. And with good reason: game development, particularly at the AAA level, has become dramatically more expensive and time-intensive over the past decade, as studios strive to ship larger, longer, and more content-rich experiences. In that environment, tools that promise faster iteration, higher output, and potential cost savings hold obvious appeal.

The risks, however, are not only reputational. This is a nascent technology whose development is outpacing regulation, and questions around IP and fair use remain murky. Although studios have always faced IP clearance and infringement questions, generative AI exacerbates existing issues while introducing new ones. The best strategic approaches to using these technologies are those that weigh the new and increased risks against the potential benefits, all while considering risk-mitigation methodologies.

Courts and regulators may take years to settle complex questions about AI and IP, while the industry moves at full speed and the court of public opinion renders instant verdicts. For studios and publishers, taking proactive steps now is the best way to avoid becoming tomorrow’s headline.

Where AI sits in the pipeline matters

AI use in game development is not uniform or easily reducible, and each use case carries its own level of legal and reputational exposure. Consider the following examples:

• Coding and backend workflows. The least visible place AI is taking hold is in development and coding. As generative AI tools reshape software production more broadly, it is no surprise that studios are increasingly using AI to support engineering work. In this respect, gaming is following the same trajectory as the wider tech industry. Reputational concerns are more manageable in this area, as these uses are largely internal, less visible, and already accepted across the tech industry. IP concerns, on the other hand, are more complicated, but generally present less exposure than at later points in the pipeline.
• Concept art and early ideation. AI is widely used in the earliest creative stages, where it can accelerate brainstorming and visual exploration. Because concept work is typically internal, iterative, and far from the final shipped product, its risks are generally more manageable, especially when teams use outputs for ideation rather than as finished materials (which reduces the likelihood of IP issues). Even here, however, studios must ensure that early designs are not implemented without appropriate review.
• Production assets. The stakes rise significantly when AI is used to generate final in-game assets like character models, textures, environmental art, animations, or voice and dialogue content. Reliance on third-party models introduces additional uncertainty: Without knowing what training data sets have been used, there is a possibility that outputs may contain content that would otherwise have to be cleared.

In short, the closer AI gets to the shipped player experience, the more important rigorous clearance and review become, by both internal QA and compliance teams and, where appropriate, IP specialists.

Third-party tools, first-party accountability

Most studios adopting generative AI are not building models from scratch. They often rely on third-party systems, such as external generative art tools, AI-driven voice solutions, and widely used developer copilots. Because these tools are largely accessible and increasingly integrated into standard production pipelines, they introduce a distinct category of risk: The more removed you are from the model, the less visibility you have into its contents.

For instance, a studio leveraging a third-party AI tool might not know what training data that system relied on, how carefully it was sourced, or what safeguards exist to prevent the reproduction of protected, harmful, or even illegal material. That uncertainty is part of what makes generative AI different from more traditional outsourcing relationships. Risk does not stay neatly upstream.

“Game developers should not assume they are immune from downstream exposure”

For example, if an output reproduces someone else’s protected work, and that work makes its way into a released game, the studio could be held responsible as the one publishing and monetising it. While most litigation to date has focused on AI providers, game developers should not assume they are immune from downstream exposure. As we see more widespread adoption and exploitation by users of these tools, it would not be unreasonable to see litigation targeting those users as well.

Some studios may go further by fine-tuning existing models on proprietary content; a few may even explore building their own systems. While doing this has its benefits, it also shifts legal and regulatory compliance risks and obligations as companies move along that spectrum, potentially introducing additional governance costs and documentation burdens as rules continue to evolve.

Copyright: familiar laws, new complexities

At the centre of much legal uncertainty about generative AI is a familiar issue: copyright. Although game studios have always had to navigate issues around copyright law, generative AI changes the scale and opacity of an organisation’s potential exposure.

The courts are unlikely to resolve that issue soon. Federal judges are currently issuing split rulings on whether training AI models on copyrighted works can qualify as fair use. Even if courts ultimately answer this question, developers should not assume that legality will automatically translate into community acceptance. Developers should also consider that rulings might not deliver any bright line rules regarding use of the technology.

Copyright concerns also cut in the opposite direction. In other words, the question is not only whether AI-generated work infringes on others, but also whether studios can successfully protect what they create with these tools. The US Copyright Office recently reaffirmed that works generated entirely by automated systems are not eligible for protection. That has significant implications for an industry built on controlling valuable creative IP. Fortunately, recent copyright registrations suggest that where human creators can demonstrate meaningful control through selection, revision, and deliberate creative decision-making, some limited form of copyright protection might persist, even if AI was used.

The bottom line? Because generative AI accelerates production but with the potential of expanded or additional risks, it increases the need for careful oversight. Studios that treat it as a shortcut around IP fundamentals are likely to find themselves exposed both to infringement claims and uncertain protection of their own creations.

What regulators may focus on next

While much of the current legal conversation has focused on copyright and training data, some of the most complicated regulatory issues may still lie ahead.

One emerging flashpoint is the intersection of AI with likeness and the right of publicity. As generative systems make it easier to replicate recognisable voices or likenesses – or allow users to embed these elements into gameplay – studios will face difficult questions about whether appropriate consents have been obtained. Legal frameworks such as publicity rights may become increasingly relevant to studios and increasingly scrutinised by regulators, as these tools blur the line between authorised use and digital replication.

“Studios will face difficult questions about whether appropriate consents have been obtained”

A related but distinct risk arises from dynamic dialogue and unscripted behaviours in AI-enabled environments. As generative systems make it easier for players to create open-ended interactions, regulators may increasingly focus on safety, moderation, and downstream misuse, particularly where studios cannot fully control what AI-driven characters say or do.

A third area of potential regulatory attention involves AI-driven engagement and monetization, particularly in live service contexts. The industry has already faced scrutiny around loot boxes, microtransactions, and the protection of younger users. Generative AI introduces the possibility of far more personalised systems, including tools that can adapt offers or progression mechanics to individual player behaviour in real time.

If AI is being used to deepen retention or encourage spending with unprecedented precision, regulators may begin to view these systems as catalysts for addiction, financial exploitation, and harm to minors. Studios experimenting with proprietary model development should also be aware that compliance obligations may vary widely across jurisdictions, particularly where regulatory frameworks require training data transparency for models placed on the market that continue to evolve.

Reputation, regulation, and the path forward

Even as lawmakers and courts work through unsettled questions, studios should not assume that risks are remote. In practice, the most immediate disincentive for AI use in games may not be regulatory at all, but reputational, in the form of a bad headline or a social media flare-up.

While games are a business, they’re also an artistic medium, supported by a community that places high value on human creativity. Regardless of what the law permits, public perception remains part of the broader business calculus for studios adopting AI.

This state of affairs is increasingly formalised through “soft regulation.” Platforms have begun introducing disclosure requirements; awards bodies and industry institutions are developing and enforcing standards; and labour agreements are incorporating AI-related clauses. As a result, studios must now review not only legal developments, but also evolving platform policies and contractual constraints.

The economic pressures driving adoption are real, and generative AI tools will only grow more entwined with game development. The studios that succeed will be those that integrate AI deliberately, with clear governance and an awareness of both legal and reputational exposure. Where is AI permitted in the pipeline? Where is it prohibited? How does a studio balance efficiency with creative integrity? In 2026, executives can’t afford to wait until a crisis hits to answer these pivotal questions.