Brandmark/Primary

No results found

AU
You are currently viewing our Australia site.
Select your preferred location for tailored content to your location.
  • Australia
  • Asia Pacific
  • New Zealand
  • Singapore
  • Malaysia
  • Hongkong
Back to News and Insights

Generative AI vs The Artist? The Changing Face of Board Game Design

Copyright

Aesthetics have been a key feature of board games throughout history. Think of intricately carved chess pieces, the black and white simplicity contrast of a Go set, the brightly coloured snakes on a Snakes and Ladders board, or the well-loved top hat, shoe, or car tokens of the original Monopoly. As the board game hobby has evolved, game theming and components (especially meeples) have driven game appeal across the spectrum, from simple card games to the heaviest of the Euro games. Gameplay alone is rarely sufficient to push a game to bestseller status. Users often invest in premium editions of games (costing many hundreds of dollars) with higher quality components to enhance and improve their gameplay experience. Board games are invariably displayed for sale in a manner that shows off the quality of their artwork.

Until the very recent past, game publishers would hire or contract artists to work on games. In many cases, their work would be a key selling feature of the game and prominently promoted as such. For example, artist Kyle Ferrin has worked on many games for studio Leder Games (including the very popular Root and Oath titles) and his artwork is widely discussed and admired by users. The publisher also credits his work and features him on its website: About – Leder Games. Popular board games often lead to spinoff “expansion” sets, and the same artists would typically be hired to work on expansions to preserve the style and theme of the original game.

However, generative AI tools such as DALL-E (https://openart.ai/) and Midjourney (Midjourney) are increasingly able to generate artworks, and to do so based on themes or styles of artworks on which they are ‘trained’. This has led to significant controversy, including as a potential threat to the livelihood of artists working in the hobby and board game industry.

As many avid observers of the games industry would be aware, the Kickstarter platform is often used to launch new board games or expansions. In 2023, Kickstarter announced a policy requiring disclosure of AI used in games and other products offered on its platform (Introducing Our AI Policy (kickstarter.com)). One of the first games disclosed as using AI-created imagery on the platform was the blockbuster heavyweight game Terraforming Mars, which disclosed use of generative AI in relation to its forthcoming Automa and Prelude-2 expansions. The publisher Stronghold Games noted that “while all the components of this game have a mix of human and AI-generated content nothing is solely generated by AI.”

Other game publishers have since adopted positions for and against the use of generative AI in game design. Recently, Wingspan publisher Stonemaier Games stated that it “does not, has not, and will not use any form of AI to replace or augment* creative work”  (Generative AI? Not for Us! – Stonemaier Games).

Where does intellectual property law stand on all of this? In many ways, it is too early to say. A number of legal proceedings are on foot internationally concerning AI including infringement claims relating to unauthorised use of works in “training” and reproductions of works created via generative AI tools. However, some non-exhaustive issues to be considered are as follows:

  • Under Australian law, copyright protection is only available to works which are the product of “independent intellectual effort”. While the issue of AI authorship has not squarely come before the Australian Courts in the copyright context, it is generally considered that this means that the author of a copyright work must be a human (as is the case in relation to the inventor of a patent). Creation of generative AI-created artworks may involve substantial effort from individual makers or feeders in terms of prompts, revisions and direct manipulation of the image output. The question of whether that effort is sufficient to warrant copyright protection for the person working with the generative AI tool will likely fall to be decided on a case by case basis;
  • Works which are created by both human authors and generative AI tools (which appears to be the case for the Terraforming Mars expansions above) risk creating co-authored works with the copyright jointly owned by a human author/artist and (depending on the Australian Courts’ approach to AI generated works) potentially part owned by the person who directed the AI creative process. There is also the possibility that the copyright work may be part owned by the person or persons involved in the creation and programming of the AI tool itself and the input used to train it, although this is an untested possibility and would need to be determined by the Courts. This potential co-authorship may complicate determination of authorship, commercial exploitation of works, and any related infringement proceedings;
  • Where generative AI is trained on the work of particular artists and directed to produce output in a similar theme or style, copyright infringement risks may be involved. If a Court was to find that training an AI on copyright material amounts to copyright infringement, it is possible that the output of this training (ie, the ultimate artwork) may be considered to be the product of copyright infringement. This issue will likely be considered by the American Courts in George R.R. Martin’s lawsuit against OpenAI;
  • Use of generative AI to produce artworks which look like or closely resemble the work of another artist or include elements of their work may present complications including in respecting the moral rights of attribution of original artists while avoiding misrepresenting generative AI-created works as those of the original artist (which would be prohibited under the Australian Consumer Law), especially where the artist has been integral to the success of the original game;
  • Commercial agreements with game artists may need to accommodate the subsequent use of the artist’s work in generative AI-created works, including training generative AI.

 The full scope of legal issues relating to use of generative AI in game design are yet to be settled. However, some of these risks may already be baked in. Stronghold Games CEO Travis Worthington commented in an interview that “I don’t think this is going to go back in the bag. It’s too powerful a technology.” (Terraforming Mars team defends AI use as Kickstarter hits $1.3 million – Polygon). As with many other industries, businesses which do not use generative AI at all may risk inefficiencies and become uncompetitive. Generative AI is also widely used in game testing. Given the importance of art to board game hobbyists and the controversy in the hobby concerning use of generative AI, the uses which will and will not be acceptable to board game consumers may end up as important to the evolving use of generative AI as the legal position.

Beautiful games much enjoyed by the author include Azul, Illusion, Patchwork (Americana), Jaipur and Takenoko. All are too old to have involved generative AI artwork. With the increasing sophistication of generative AI art engines and the use of combined generative AI and human inputs, players may not be able to tell the difference between games without generative AI input and those with such input other than via their publication date or information from publishers.

 

This article forms part of DCC’s Gaming & IP initiative