Epic's next Disney game could become a very public test of how far AI can carry a major production. Reporting published on October 12, and picked up across the gaming press today, says the studio has told developers on an unannounced Marvel and Star Wars shooter to adopt an “AI-first” approach. The game has also been delayed.

The interesting part is how firm those reported instructions are. Plenty of developers choose AI tools for particular tasks. Telling a team that code must come through those tools changes who gets to decide how the work is done. Epic will eventually have to demonstrate that the change helps the people making the game, as well as the schedule.

The Instructions Go Beyond Trying a Coding Assistant

According to Ars Technica's account of Bloomberg's reporting, the project involves Claude-generated work managed by developers. A director reportedly said, “No more hand-written code,” and that art would be “generated with full AI workflows.” The reported March 2027 target has slipped, with no replacement date communicated.

The distinction between a preferred workflow and an absolute rule matters here. Creative leadership requires people to judge the output and decide how to improve it. How much discretion would those people retain over the method?

Would an engineer be free to rewrite a troublesome system directly? Could an artist finish a piece using whichever method produces the best result? Those are practical questions Epic could answer. A production policy becomes much easier to assess once the people working under it know what happens when the preferred tool gets in the way.

Sweeney's Promise Is More Games, With People Still Involved

In an October 12 social media statement reproduced by GIGAZINE, Tim Sweeney described a substantial reorganisation around AI. He said the aim was to let Epic's developers build more and better work, rather than produce the same amount with fewer people, and to extend the benefits to teams using Unreal.

That is a worthwhile ambition. If a small team can test a mechanic sooner, or spend less time on repetitive implementation, it has more room to develop the parts players will notice. Giving a developer time to try a second version of an idea can be more valuable than getting the first version out quickly. Speed matters when it buys that opportunity.

But the choice of measurement matters too. Counting generated code or completed assets would tell us relatively little about whether a game is improving. We'd want to know whether the team can integrate the work, revise it without breaking something else, and keep the result running well. Epic hasn't presented those project-level results in the material reviewed for this article.

The Disney Name Makes This a Particularly Visible Experiment

Disney's February 2024 announcement paired a planned $1.5 billion equity investment in Epic with a multiyear collaboration. It described a persistent entertainment universe connected to Fortnite, powered by Unreal Engine, with experiences drawing on Disney, Pixar, Marvel, Star Wars and other properties. That broader announcement shouldn't be treated as a detailed description of the specific shooter now being reported.

It does explain why the production approach will attract attention. Familiar characters give players expectations before they've seen a single match. If the animation feels wrong or a character looks oddly inconsistent, the audience will have plenty to compare it with. The franchises are an advantage for attracting interest, and a demanding reference point for the finished work.

There is a useful contrast with Yuji Horii's small AI-made browser experiment we covered yesterday. A quick personal project lets someone explore an idea at modest scale. A major licensed shooter has to sustain that idea through a much larger production. Success in the first setting would give us a reason to experiment in the second, but it wouldn't settle how the second should be managed.

A Delay Gives Epic Something Concrete to Prove

There is also an uncomfortable employment backdrop. In its March 24 announcement, Epic said it was laying off over 1,000 employees after declining Fortnite engagement put pressure on its finances. Sweeney explicitly said those cuts weren't related to AI. That statement belongs in the discussion; the later production pivot doesn't establish that AI caused the earlier layoffs.

For the delayed game, the sensible test is whether the new approach produces a better build and a credible path to release. Developers should have enough freedom to fix what fails, including the freedom to decide that a particular task needs a different method. Otherwise, “AI-first” risks becoming a rule the team has to satisfy alongside all the work it already needs to finish.

We'd like to see Epic show the results. A convincing playable game would be far more useful to this debate than another argument over how much of its code came from a model. Until then, the reported mandate is a production experiment whose outcome remains open.

Sourcing note: the project details are attributed to Bloomberg's reporting as reproduced by Ars Technica. We also consulted GIGAZINE's reproduction of Sweeney's statement, Disney's original partnership announcement and Epic's March layoff notice. We haven't inspected the internal messages or played the unannounced game.

Would you trust an AI-first production rule on a project this size, or would you want developers choosing the tools task by task? Let us know your thoughts in the comments. And if you've got a game you'd like us to review, send it over to editor@gamevibe.io.