Can a Game Studio Sell or License Its Data to AI?
AITrainingDataBroker.com
Can a Game Studio
Sell or License
Its Data to AI?
Its Data to AI?
What a studio keeps after launch, and what an AI lab may want to license
Jimmy Lin
Can a Game Studio Sell or License Its Data to AI?
A game that runs for years leaves a trail behind it: every bug report, every design change, every build. So is that trail worth anything to an AI lab? Let us look at what a studio keeps.
The video
Full script
A game that runs for years leaves a trail behind it: every bug report, every design change, every build. So is that trail worth anything to an AI lab? Let us look at what a studio keeps.
Most owners think the only thing a studio could ever license is the game itself. But it is actually only the finished level. Behind it sit years of tickets, design notes and source history, and that trail of how the game was built, broken and fixed is the kind of data an AI lab may want to license. It also comes with rules about who wrote it and what players agreed to, and we will go through them. This is general information, not legal advice. First, let me tell you who I am.
I'm Jimmy. I have worked in data at Fortune 500 companies, as a data analyst, a data engineer and a software engineer, and I am trained in data science. I am also a Canadian military veteran, specialized in data analysis and reporting.
If you want to sell your data, I am your broker and technical advisor. I bring your data to several AI companies, and I guide your team through the technical side, including compliance. AI companies pay my fee.
Already shopping your data around? Contact us anyway. We can still work through the terms with you and help you negotiate a better price.
If you are buying data for an AI lab, I find the right data and see it through to delivery, so you can focus on AI research.
Let us follow one bug through a made-up studio, say a door that does not open. A player reports it, the team decides how urgent it is, a developer fixes it, a tester checks the fix, and one line goes into the release notes. Every step leaves a written note, and together the notes show how a problem was found, argued over and solved. That whole path is the data, not only the last line players see. By the way, we built a free data value calculator, and the link is in the description. But does anyone actually study this kind of data? Yes.
Now, here is the research. According to a research paper on arXiv, from March 2021 by Truelove and colleagues, researchers read the update notes of 30 popular PC games, 12,122 bug fixes in all, and sorted them by type. Crashes were the bug type fixed most often in urgent updates. And according to another paper on arXiv, from October 2022 by Taesiri and colleagues, AI models were given descriptions of events in gameplay clips and asked which event was the bug, and on one test they picked correctly about 71% of the time. So both public update notes and gameplay clips are already being studied, and a studio's own tickets hold the same kind of material.
Next, the source code, which is often the biggest piece. A studio's code is not often all its own, because it sits on top of an engine, outside libraries and open-source parts. According to a research paper on arXiv, from November 2022 by Kocetkov and colleagues, a team building a code dataset kept only permissively licensed code, 3.1 terabytes of it across 30 programming languages, and let developers ask to have their code removed. In plain words, AI teams look at the license of every part. So the lines your team wrote can be considered, but the parts that belong to others cannot be licensed by you.
Then comes the question every credits screen answers: who made it? According to the US Copyright Act of 1976, work an employee makes as part of the job usually belongs to the employer. A contractor's work depends on the written contract. Licensed music and art were only licensed to you, so you can not license them on, and player mods and creations depend on the terms players accepted. So one studio can hold work with 4 different owners, and each one changes what can be licensed.
Now the gameplay telemetry, which is the log of what players did. It looks harmless, but a player ID, an internet address or a device number can point to a real person. The EU privacy law GDPR, from April 2016, says so: it names internet addresses and cookie IDs as online identifiers that can identify people. So a telemetry log is de-identified before an AI lab sees it. We remove the IDs, the addresses and the chat, and keep the pattern of play.
But some of a game's data should not be part of any license at all. In December 2022, the US Federal Trade Commission announced a record penalty against a popular online game over children's privacy. Its order also bars turning on voice and text chat for children and teens without clear consent. So we leave these out of the license: data from players under 13 without a parent's consent, chat and voice with no consent, private messages between players, and anything a contract bans. These are not part of what the AI lab receives.
Of course, this will not fit every studio, and I would rather say so now. A game that launched last month has no years of tickets to show. A very small team usually has too little data to license, and a studio that made its work for a publisher may not own it. But if your studio has run a game for years, with a team of 30 or more people, keep watching, because here is what to ask yourself.
Here are 4 questions to ask about your own studio. How many years of tickets do you keep? Who wrote the code and made the art? What did players agree to in your privacy policy? And does a publisher hold any rights? Answer those 4, and you know most of what we would ask first.
So what happens next? First, we de-identify a sample of your data. As your broker, I take that sample to several AI labs and bring back their offers, so you can compare them and choose the best price and terms. As your technical advisor, I guide your team through the export, the format and compliance. You make the final call, and we move forward only after your team signs off.
A game is more than its last build, and the tickets, notes and logs your studio already keeps may be worth something to an AI lab.
If you run a company with 30 or more people, the data you already have could become a new revenue stream.
Want to find out more? You can visit AITrainingDataBroker.com to learn how licensing your data works.
If you have further questions, leave them in the comments, and I will answer them in the next video.
Thank you for watching.
I will see you next time.
More slides
AITrainingDataBroker.com
Will Synthetic
Data Replace
Your Records?
Your Records?
What synthetic data can do, what it can not, and why real data still counts
Jimmy Lin
AITrainingDataBroker.com
Can a Chip
Company License
Its Designs?
Its Designs?
What a design folder holds, who owns it, and which files cannot be licensed
Jimmy Lin
AITrainingDataBroker.com
Sell Egocentric
Video
to AI Labs?
to AI Labs?
What it is, who wants it, and who has to agree before it can be licensed
Jimmy Lin
