Epic Games is making a much bigger bet on artificial intelligence than simply adding another AI tool to its development toolbox.
According to reports, the company has shifted an unannounced Disney collaboration toward an AI-first production model, with the team being instructed to use AI heavily for both code and artwork. The project, reportedly a multiplayer shooter featuring Marvel and Star Wars characters, has also been delayed beyond its previously targeted March 2027 release. Bloomberg
That makes this story more interesting than another debate about whether developers should use AI.
Epic is effectively testing a different question:
What happens when AI stops being an assistant and becomes part of the foundation of game production?
Quick Summary
- Epic’s Disney collaboration has reportedly moved toward an AI-first development model.
- The unannounced shooter was previously targeting March 2027 but now has no announced replacement date.
- Reports say AI will be used extensively for code and art production.
- Developers are still expected to guide, review and control AI-generated work.
- Epic says its goal is to increase what developers can create rather than simply replace them.
- The project could become an important test for AI-driven game development at a major studio.
- The biggest questions are still quality, consistency, originality and how much human creative control remains.
Epic Has Reportedly Changed How the Game Is Being Built
The reported change is unusually broad.
Developers working on the Disney project were told to adopt an “AI-first” production model, with internal communication reportedly calling for an end to manually written code and full AI workflows for artwork. Tools including Anthropic’s Claude are part of the reported process, with human developers directing the generated output. Bloomberg
That distinction matters.
AI-assisted development usually means a developer remains at the center of the process. A programmer might ask an AI system to explain a bug, generate a function or suggest an optimization, then decide what belongs in the final game.
An AI-first workflow goes further.
The starting assumption changes from “How can AI help the developer?” to “How much of the production process can begin with AI?”
That is a much bigger experiment.
AI-Generated Code Is Not the Same as AI-Assisted Coding
Game code has an awkward relationship with generative AI.
A model can produce working code surprisingly quickly. It can also create something that looks perfectly reasonable while introducing problems that only become obvious later.
Games make that problem harder.
A single feature can touch gameplay logic, animation, physics, networking, UI, audio and backend systems. A small change in one area can create an unexpected problem somewhere else.
Human programmers therefore still have an important job even when AI generates much of the initial implementation.
They need to decide:
- What the system should actually do
- Whether the generated code is correct
- How it interacts with existing systems
- How performance is affected
- How bugs should be diagnosed
- Which generated solutions should never reach production
Epic’s own description of its approach emphasizes developers as the people guiding the process rather than simply removing humans from it. Ars Technica
That distinction could become one of the most important parts of this experiment.
Art Creates a Different Problem
Code can often be tested against a clearly defined result.
Art is much harder to judge that way.
A generated texture might technically work. A character model might technically fit the specification. An environment might contain everything requested by an artist.
None of that guarantees that the result has a consistent visual identity.
A large game can contain thousands of assets. Characters, weapons, environments, effects and interface elements all need to feel as though they belong to the same world.
Marvel and Star Wars make that challenge even more significant.
These are recognizable universes with established visual languages. A technically impressive asset can still feel wrong if it does not fit the style, tone or design rules of the franchise.
That leaves human artists with a different role.
Instead of creating every asset from the beginning, they could increasingly become directors, editors and quality controllers of generated material.
Whether that produces better results remains unproven.
The Real Test Is Not Speed
AI can make certain production tasks faster.
That part is not particularly surprising.
The harder question is what happens after the first draft.
Games are rarely difficult because someone cannot produce a first version of an asset or system. The difficult part is refining that work until everything functions together.
A generated weapon can fire.
Making it feel right is another problem.
A generated environment can look impressive.
Making it support gameplay, performance requirements and a coherent art direction is another.
A generated piece of code can compile.
Making it stable inside a huge interconnected game is another.
That is why the reported delay is particularly interesting.
Epic is reportedly introducing its AI-first approach while the project is already behind its previous schedule. Bloomberg
AI therefore isn’t simply being tested on a blank project.
It is being introduced into a production that already has complexity, deadlines and expectations.
Could Smaller Teams Benefit From This?
This may be the most important long-term question.
Large studios have traditionally needed enormous teams because modern games require specialists across programming, animation, environment art, character art, audio, design, QA and production.
If AI genuinely allows a small team to accomplish work that previously required a much larger group, the impact could extend far beyond Epic.
A five-person studio might prototype systems faster.
An independent artist could create more variations.
A small programming team could spend less time on repetitive implementation.
But there is a difference between making more material and making a better game.
If AI lowers the cost of producing assets, the industry could simply end up with more content competing for players’ attention.
Quantity is not the same thing as creativity.
Unreal Engine Developers Will Be Watching
Epic’s position makes this experiment especially significant.
The company does not only make games. It also develops Unreal Engine, which is used by studios around the world.
Epic CEO Tim Sweeney has described the company’s broader AI push as a way to let developers produce more and better work and eventually extend those benefits across teams using Unreal. Ars Technica
If the Disney project demonstrates that AI-first production can maintain quality while increasing development capacity, other studios could take notice.
If the experiment struggles with quality, consistency or production complexity, that lesson could be just as valuable.
Either way, Epic is creating a case study the wider development industry will be watching.
The Biggest Risk Is Losing the Human Filter
There is a tempting assumption that AI automatically makes development more efficient.
It does not.
AI can generate enormous amounts of material very quickly. Someone still has to determine what deserves to survive.
That human filter becomes more important as the amount of generated material increases.
Without strong direction, a studio could end up spending less time creating assets but more time sorting through mediocre ones.
That would change the bottleneck rather than remove it.
For a game built around two of entertainment’s biggest fictional universes, the standard for that final human judgment will be extremely high.
Epic’s Experiment Could Change Game Development
Epic’s AI-first shift is important because it moves the discussion beyond individual AI tools.
The company is reportedly experimenting with an entire production philosophy.
Code generation.
Art generation.
Human supervision.
Production workflows.
And potentially, a different relationship between developers and the tools they use every day.
The result could eventually show that smaller teams can create larger games, that repetitive development work can be reduced, or that certain parts of production simply work better with humans at the center.
Right now, none of those outcomes is guaranteed.
The Disney project does not have a new public release date, and there is not enough information to judge the finished game because the game itself has not been shown publicly. Bloomberg
That is what makes the project worth watching.
Epic isn’t just making a game with AI. It may be testing how a major game is made in the first place.
Frequently Asked Questions
What is Epic’s AI-first game project?
It is an unannounced Disney collaboration reportedly featuring Marvel and Star Wars characters. Reports say Epic is rebuilding its development workflow around extensive AI use.
Is Epic’s Disney game delayed?
Yes. Reports say the project was previously targeting March 2027 but has now been delayed without a new public release date. Bloomberg
Will AI create the entire game?
Reports indicate that AI will be heavily involved in code and artwork production, while human developers will guide and manage the generated work. That does not mean the entire game is being created autonomously.
Why is Epic using AI for game development?
Epic says its broader AI strategy is intended to increase what developers can create and improve productivity rather than simply reduce the number of people working on games. Ars Technica
Could AI-first development change the gaming industry?
Potentially. If Epic can demonstrate that AI-first workflows maintain quality while increasing development capacity, other studios may adopt similar approaches. The finished results will matter more than the technology itself.