AI coding tools have turned retro game porting from a years-long craft into a weekend project, and the communities that built the craft are paying for it. The Verge AI reports that over the past month, viral ports have appeared at a remarkable pace: Halo in a web browser, Wind Waker on the iPhone, Super Smash Bros. in mixed reality. Last week, a community website that tracked decompilation projects shut down, and its owner blamed AI directly.
This story is about more than old games. It’s an early, very visible case of what happens when AI makes skilled labor cheap inside an open-source community that runs on trust and credit.
🕹️ What Decomps Are and Why AI Changed Them
Decomps and recomps (short for decompile and recompile) are reverse engineering. You break a game down to its building blocks, then rebuild it so it runs on hardware it was never made for. Done by hand, it’s slow. One group spent years decompiling the original Animal Crossing for the GameCube.
AI tools have cut that timeline down sharply. Chris Sotraidis, an AI consultant who ports games as a hobby, went from one project to a dozen in a few weeks, including a mobile port of Ocarina of Time built on an existing community decomp. He’s candid about it: “the pace and volume would not have been possible for me without it.” He also offered a line that sums up the whole shift: “The tools allow you to move much faster if you know what you want.”
⚔️ Where the Community Breaks
The trouble isn’t the speed itself. It’s who benefits from it. The site’s owner, writing as Encounter, admitted that “with the right guidance, human input, and code reviews, AI can expedite the decompilation process in a significant way.” But Encounter also said “the community aspect of decompilation has been broken.”
According to The Verge AI, the abuse comes in a few forms:
- AI “finishing”: people take someone else’s unfinished decomp and use AI to complete it
- Port racing: people rush out ports of decomps they never contributed to
- Hostile forks: copies that, in Encounter’s words, “put little to no effort into meaningfully contributing to existing projects”
There’s a deeper preservation cost too. Decomps exist partly so people can understand why a game was built the way it was. When AI does the work, that understanding never lands in anyone’s head. You can drop Spider-Man into a Batman game, but you won’t learn the programming decisions that matter for faithful restoration.
🔍 Two Camps, Both Partly Right
The stakes are real. Over 90 percent of classic games are estimated to be “critically endangered,” so anything that pulls new people into preservation has value. AI has turned “decomp” from niche jargon into headline material and lets non-programmers try out ideas.
Sotraidis argues the quality will rise: “a lot of what gets dismissed as slop today can improve as the tools get better.” What stands out is his next point, which concedes the preservationists’ main complaint: “Better tools aren’t automatically going to make people give proper credit or treat each other well.” That’s the real issue. Model quality won’t fix a norms problem.
The industry isn’t waiting for that debate to settle. Epic Games launched a vibe-coded version of its PC store and has reportedly moved to “AI-accelerated development” for an upcoming Disney collaboration. Tim Sweeney says AI will bring “a new golden age of gaming.” The Verge AI also notes that Xbox CEO Asha Sharma has moved to embrace the decomp scene. Anyone hoping to keep AI out of this space is probably fighting a losing battle.
🧭 Why This Matters Beyond Gaming
Retro decomps are a stress test for every open-source community. The pattern is likely to repeat elsewhere:
- AI lowers the cost of building on top of someone else’s work close to zero
- Credit and contribution norms stay exactly the same
- The people doing the slow foundational work burn out and walk away
When those people leave, the shortcut-takers lose the foundation they were building on. A shut-down resource hub hurts the AI porters too.
✅ What Practitioners Should Do
If you’re building with AI on community-made code, in gaming or anywhere else:
- Contribute upstream. Send fixes back to the original project instead of forking and racing it.
- Credit loudly. Name the decomp or library you built on, right at the top of your release.
- Document what the AI did. Write down what you learned so the knowledge survives, not only the binary.
- Maintainers, set the rules now. Clear AI-contribution policies and attribution requirements beat shutting the doors later.
My recommendation: treat AI speed as a reason to invest more in community, not less. The projects that last will be the ones where fast tools and slow understanding work together. Full details are in the original report from The Verge AI.