On 10 January 2024, Valve published a statement updating its stance on the use of AI generated content in video games that are distributed via Steam. This update is important for video game developers and publishers who may want to distribute their games via Steam where they: (a) have AI generated content in their upcoming games already; or (b) are considering having AI generated content in their upcoming games.

Valve’s stance on AI generated content prior to this update was slightly stricter, with many in the industry reporting that Valve had introduced a ban on games being distributed via Steam where they contained AI generated content. In their most recent statement, Valve has put this down to needing more time to learn about the “fast-moving and legally murky space of AI technology”, as Steam described it. Having taken a few months to consider their options and learn more about AI, Valve has seemingly confirmed its stance on AI generated content in games for the foreseeable future.

The main points to note from Valve’s statement are as follows:

• The Content Survey which developers fill out when submitting their game to Steam will now include a section for “AI Disclosure”. Here, Developers will need to describe how AI is used in the development and execution of their game.

• The section on AI Disclosure in the Content Survey will split AI usage in games into two broad categories, both of which will need to be addressed (where applicable). These categories are Pre-Generated AI and Live-Generated AI. A developer will have certain responsibilities depending upon which (if any) types of AI are being used in their games. This is discussed in more detail below.

• The section on AI Disclosure in the Content Survey (or significant parts of it) will be used on the Steam store page for the game to which it relates. The logic behind this is so that Valve’s customers can also understand clearly how the game they are purchasing utilises and implements AI technology. We give further thoughts on this under “Publication of Disclosure on Steam Store” below.

Pre-Generated AI:

Pre-Generated AI content includes any content in a game (art, code, sound) which is created via AI tools as part of development process.

When a developer signs up to Steam’s Distribution Agreement, they are warranting (promising) that their game, amongst other things, won’t infringe any third-party’s intellectual property rights. Valve have confirmed that in their pre-release review of a game, they will evaluate the output of AI generated content in the game with the purpose of ensuring that the game meets the warranties that have been provided by the developer in their standard Distribution Agreement.

This is clearly a response to the inherent legal risks that generative AI poses from an intellectual property perspective. To understand the legal risks of generative AI, we must first explore how it works.

Firstly, we need to understand “text and data mining”. Text and data mining is a technique that involves analysing substantial “datasets” to identify patterns or trends. The product of text and data mining has a number of uses, but one of the most prominent at the moment is use as input and training data for AI systems and models.

The datasets used for text and data mining are vast and whilst they can be populated with whatever the creator of the dataset desires, they are often full of data that either is unlicenced (meaning permission has not been obtained from the owner of copyright in the data) or that has been “scraped” from the internet using a process known as “web scraping”. Web scraping is a data collection method that operates by searching the internet (including selected websites) and then extracting data that is deemed useful in some way. The extracted data can then be copied, retained in large datasets and used for things such as data analysis/mining.

One of the main issues in using generative AI is that ”end users” of AI software do not know how the AI software has arrived at the “end product” that is provided to them (and what data has been used to generate that end product). If the AI software provider has used unlicenced materials (which are subject to copyright) in their datasets and those datasets are then utilised to generate the end product, then the owner of the AI software is potentially committing copyright infringement. If the end user then went on to use the end product in their games, they may also unknowingly be committing copyright infringement.

Live-Generated AI:

Live-Generated AI includes any type of content that is created with the help/assistance of AI tools whilst the game is running. This is slightly different to pre-generated AI in the sense that it envisages AI software operating in the background whilst the game is being played, rather than AI software just being used to create the game itself.

In addition to following the rules on Pre-Generated AI (as set out above), developers who have games that implement live-generated AI will also need to disclose in the Content Survey what type of “guardrails” are being put on the AI software they are using, to ensure that the software isn’t generating illegal content.

This seems to be more of a tailored response to the risks that including live-generated AI in games poses. Examples of live-generated AI could include using AI software to allow players to have a unique experience during the course of gameplay through the operation of generative AI software built into the game.

In those scenarios, there is of course still the risk of intellectual property infringement from use of the AI software (in relation to output from the AI software potentially infringing a third party’s rights). However, there are also other risks involved which potentially aren’t present when only using pre-generated AI. Examples include misuse of such AI software to facilitate the generation of harmful, obscene, immoral or illegal “output” content.

As a second safeguard for live-generated AI, Valve have confirmed that Steam will introduce a new system that allows players to report illegal content inside games that contain live-generated AI. This will be done by an in-game overlay that players can use to submit a report to Valve where they feel that they have encountered content that should have been caught by appropriate safeguards on AI generation.

Whilst not stated within Steam’s statement of 10 January 2024, these guardrails referred to and the facility to report illegal content may be part of a framework from Steam to assist with its compliance with the UK’s Online Safety Act 2023.

Publication of Disclosure on Steam Store

Given that Valve’s intention is to publish the majority of the AI Disclosure section in the Content Survey on the Steam store page for the game to which it relates, developers will need to be particularly cautious about what information they include. Importantly, they will need to strike a balance between explaining the relevant use of AI effectively and keeping the content “user-friendly” (so that an average customer could understand what is being said). Transparency as to use of AI appears to be Valve’s goal here, and developers should bear this in mind when submitting their Content Survey.

Concluding

Valve’s changes to their stance on AI generated content in games should result in them being more open to releasing games that use AI technology via Steam. That said, developers and publishers need to familiarise themselves with these new rules to ensure that they are complying with Valve’s policies where their game contains AI-generated content.

If you need legal advice on any of the topics discussed in this article, please contact me at johncotterill@eatonsmith.co.uk