AI and Copyright: Video Games Europe Position Paper
Executive Summary
Video Games Europe supports the responsible development of AI technologies and a thriving and robust AI economy. As with many technological advances, the opportunities presented by the continuing development of AI systems are tremendous for consumers of video games products and services, but come with challenges and new legal questions surrounding both the copyrightability of works produced with assistance from AI systems and the use of copyrighted works as inputs.
The video games industry and its overall value chain rely inherently on both advances in technology and an effective copyright regime to allow creativity and investment in new works to be sustained over the longer term. Copyright law has been carefully scoped to achieve this balance and includes exceptions and limitations to permit rightsholders to prevent the copying of their works while allowing new ideas and concepts to develop. We believe that it is critical that the underlying goals, purposes and balance of the existing copyright regime are upheld to support innovation and to protect the rights of creators. This balance is, we believe, reflected in Articles 3 and 4 of the DSM Copyright Directive (Directive 2019/790).
AI applications in video games do not encroach on fundamental rights or the safety of individuals. We believe that the regulation of both generative and non-generative (i.e., analytic) AI should take a risk-based approach, where the sorts of uses in video games should be considered the lowest risk and subject to the least restrictive transparency, disclosure and reporting requirements. We also believe that transparency obligations must be reasonable and proportionate, and should take into account the protection of trade secrets.
We also believe that where foundation models (i.e., large AI models trained on enormous quantities of unlabeled data) are developed and used exclusively in internal, non-high-risk settings (i.e., not available to the public nor placed on the market), used for example to generate short pieces of dialogue in an open-world game, transparency and disclosure obligations should definitely not apply.
We do not believe that creative works should be burdened with labelling obligations in contexts where users already expect to interact with AI-assisted and AI-generated content, such as in video games. To demand otherwise could be highly disruptive to the user’s in-game experience. Concerns over synthetic media and fraud, misinformation, invasion of privacy and other harms are not present in expressive works for entertainment that depict fictional worlds, such as video games.
AI technologies and how video games companies use them are still evolving, and until the surrounding issues have come into much clearer focus, we would encourage policy makers to continue to engage with industry stakeholders and to proceed with caution before making or recommending changes to either law, regulations or policy.