When sharing your game becomes a risk

There is something about the recent AI game decompilation stories that bothers me, and I am still trying to work out what I think about it.

I use AI to make games. It helps me try ideas I would otherwise struggle to find time for. I want more people to have that chance.

But the same tools can help someone take a game another person spent months or years developing, copy it, and offer a cheaper version. I can be excited about what these tools let me build and worried about what someone else could do with the result.

The part that keeps coming back to me is the demo. How do you get people interested in a game if showing it to them might help someone beat you to your own launch?

A demo used to feel like an invitation

Alliance Arts recently replaced a planned public demo for Drunken Goddess Reflux with closed play sessions. One reason was concern that someone could use AI or other tools to copy its presentation and gameplay, then release that copy before the original game. There were also concerns about managing restrictions on videos and streams. GamesRadar covered the decision, linking to the developer's announcement.

That does not prove a copy would have ruined the game's sales. It does show that the fear is already affecting what players get to try.

That is a miserable decision to have to make. You want feedback. You want people to play your game, talk about it, and decide whether they want the full version. Now you also have to think about whether that invitation gives someone else a head start.

Copying is more than one problem

I need to separate a few things here, because I initially mixed them together.

Decompilation means recovering source-like code from a compiled program. Removing a game's payment protection is a different task. Watching its gameplay and building something similar is another. A copied game does not necessarily involve decompilation at all.

Those differences matter. So does the purpose. Studying how an old game works, repairing it, or helping preserve it is very different from repackaging someone else's work to collect their sales.

I also should not jump from impressive examples to assuming every game can now be copied overnight. In Maurice Heumann's account of an AI-assisted shooter decompilation, the work took about three months. Early progress hid serious errors that needed a better way to check the results.

My concern is what happens as the time, cost, and expertise needed to produce a convincing copy go down. An exact reconstruction might not even be necessary if the result is close enough to attract buyers.

Someone else gets paid

I have usually thought about piracy this way: someone who could never afford my game was not necessarily a customer I lost. I would rather they eventually became someone who could support my work. But I cannot turn that into a rule that piracy never affects sales. I do not know what every player would have done.

A competing seller bothers me in a different way.

Imagine spending a year finding out whether a game idea works. You try mechanics, throw some away, test the pacing, and keep changing things until the game is worth playing. Then someone uses that finished work as the starting point for a cheaper substitute.

They still have work to do. But they have your result to aim at. You paid for experiments they can skip.

Some players will care who made the original. Others may only see two similar games and choose the cheaper one. Some may not even know there is an original to look for.

How often does that become a lost sale? I do not have a number. But "your real fans will support you" does not fully answer the concern. A small developer also needs to reach people who are not fans yet.

Would cheaper games solve it?

Part of me wonders whether this pressure could push game prices down. As a player, I would welcome more affordable games.

But the person making a copy can start with a lower bill. Lowering the original game's price does not necessarily remove that advantage. Even if AI reduces the cost of development for both people, someone still has to find the game worth making first.

I also caught myself thinking that fewer games might be good for an overcrowded market. That is too easy a conclusion. Which developers would leave? The ones making poor games, or the ones who cannot afford to keep going?

I would hate to see small, unusual projects disappear while companies with more money can afford to defend theirs.

Can we at least make copying take longer?

Before jumping to cloud gaming, I want to know what we could try while still letting people download and play our games.

The goal would be to buy time. I would not trust a promise that a game running on someone else's computer can never be taken apart. OWASP's guidance on resisting reverse engineering makes the same distinction: protections can raise the effort required, but cannot guarantee success against someone with enough time and access. That guidance concerns mobile apps, so it does not tell us how many days a particular game would gain against AI tools.

The first thing I would check is what the demo actually contains. If it includes the whole game with a menu check blocking the later levels, removing that check may expose content I never intended to give away. A smaller build could leave out those levels, assets, and code that only the full game needs. Someone can still copy the demo's mechanics, but they cannot extract content that was never in the download. Preparing and checking that separate build is extra work too.

Then there is the information shipped with the program. Debug symbols, which map compiled code back to names and source locations, can make analysis easier. I would check that private development files stay out of the public build and keep the files I need for crash reports separately. That removes helpful clues. It does not make the program unreadable.

Obfuscation is another option: changing code to make its purpose harder to understand while keeping its behavior. OWASP describes techniques such as renaming things and making the flow of execution harder to follow. I would want to try it on selected parts before applying it everywhere. Does it slow analysis enough to matter? What does it cost in performance, build time, and debugging? I do not have measurements that answer those questions for my games or current AI tools.

Encrypting packaged assets also has a limit. If the game needs to read them on the player's machine, the data must become readable there at some point. Encryption might complicate extraction, but I would not treat it as a safe place to hide a secret forever.

There is a more uncomfortable limit to all of this: someone making a similar game from videos does not need to understand my code. Protection against decompilation does not protect the idea visible on screen.

That makes me think the useful experiment would be to compare protected and unprotected builds on a specific task, such as recovering one gameplay system. Measure the time, human help, and AI usage needed, then compare that with the cost to development and the player. A failed attempt within a fixed budget would show a delay under those conditions, not prove the game is safe.

I would be interested in protection that buys meaningful time without making the game worse to own or play. I would be much less interested if the honest buyer gets crashes, slower loading, or an internet requirement while the copy eventually removes those restrictions. Developers need evidence for that trade too.

And then we get more cloud gaming

This is where my thoughts get more speculative.

If handing players the program makes it easier to copy, I can imagine developers wanting to keep more of it on their own servers. In the extreme version, the game runs remotely and your device sends input and receives video.

That would make the running program harder to obtain directly. Someone could still imitate what they see, though. It would not make copying disappear.

It would also mean I depend on someone else's servers to play. I want to keep playing a game after its developer moves on. I want an offline game to work without an internet connection.

I am not predicting that every developer will choose streaming. I worry that faster copying gives them another reason to consider it. Players could end up paying for that response through fewer demos, less open development, and more games that need a server to stay alive.

Shit. I like what AI helps me make. I really do not want this to be how we get more cloud gaming.

I still want to share what I am building. I want people to try it, and I want other developers to talk about their experiments too. I do not have a neat answer for how to protect that. But if showing your work starts to feel like giving someone else a product to sell, I understand why people would stop showing it.