The game industry's pipeline is having a key process quietly rewritten by Epic Games. According to reports, the maker of Unreal Engine has required an unannounced Disney collaboration shooter game to fully shift to a development model centered on artificial intelligence, with code written using AI tools such as Anthropic's Claude, and game art generated through a complete AI workflow, while human developers retreat into roles of guidance, management, and review. This means generative AI has been pushed from its former supporting role of cheering from the sidelines onto the center stage as the primary means of production for a large game project.
Bloomberg has revealed that this game is being created jointly by Epic and Walt Disney, and it is expected to blend characters and content from the Marvel and Star Wars franchises. It was originally targeting a March (3) 2027 release, but has now been postponed indefinitely, with no new date yet announced. The project's development progress has fallen behind expectations, and Epic's earlier estimates of the required headcount and timeline were also overly optimistic. It is against this backdrop that the company decided to switch to a different production method, pushing AI to the forefront to shore up its shortcomings.
A project lead put it rather bluntly in an internal message: going forward, code will no longer be written by hand, and game art will be generated through a complete AI workflow. That statement carries far more weight than AI-assisted programming in the ordinary sense—it intends for artificial intelligence to run through multiple core stages of both code and art. Epic spokesperson Jake Jones subsequently stepped in to set the tone, saying the company is using AI to accelerate development, but that the entire process is still led by Epic developers, who are the core drivers of creativity and direction, with AI merely helping people realize ideas faster rather than replacing humans. CEO Tim Sweeney also confirmed on social platforms that the company is restructuring developers on a large scale to make fuller use of AI, with the goal not being for fewer people to do the same work, but for everyone to create more and better products, and to extend this AI capability to other teams using Unreal Engine.
This pivot stirred no small tremor internally. Some developers working on the project were surprised by the new requirements and raised question marks over the quality control of AI output and the actual process: if AI-written code has errors, how much time will people spend checking and fixing it? And if AI-generated art doesn't match the overall style, how should it be adjusted? These questions directly bear on whether the new model actually saves time. From a technical perspective, AI-generated code does not mean it can ship directly—large games involve physics systems, character behavior, network synchronization, graphics rendering, and game logic, with dependencies between modules intricately entangled; even if AI writes an individual feature at high speed, humans still have to verify correctness, performance, and compatibility. The same applies on the art side: generative AI can quickly produce concept art and scene assets, but a commercial blockbuster must maintain a unified style—characters, animation, materials, and scenes must be consistent, and assets must conform to engine specifications and be optimized before they run smoothly. In other words, the speed at which AI produces work does not necessarily mean the whole project speeds up; if the team spends a great deal of time filtering, revising, and integrating, the saved production time may be eaten up by later rework—especially when the project carries mature, high-standard IPs like Marvel and Star Wars, where any substandard content may have to be redone.
Epic's AI fixation is not a passing whim. Over the past few years Sweeney has consistently been optimistic about generative AI's impact on games; he envisions a future where AI lets developers create massive numbers of characters, complex interactions, and vast worlds without having to hand-write large amounts of fixed logic for each type of interaction as in the past; he also believes small teams can use AI to make larger-scale, more richly interactive games. As the helmsman of Unreal Engine, everything Epic says and does in the industry has an amplification effect—if it truly welds a mature AI pipeline into the engine, other studios will likely copy the pattern. So this Disney collaboration game is both a scheduling experiment for a single project and an important template for Epic to test a new generation of game production systems.
But real-world doubts are hitting the company's sensitive nerve. This March (3), Epic had just laid off more than 1000 people, citing declining Fortnite user engagement and spending exceeding revenue as reasons; now, with news of fully pushing AI development coming one after another, outsiders naturally ask whether AI is helping existing employees become more efficient, or will ultimately cut the manpower needed for game development. Sweeney previously denied that the March (3) layoffs were caused by AI replacement, stressing that they were to cut costs and restore financial stability, and this time he again reiterated that AI is not about having fewer people do the same work. The company has not yet explained how this adjustment will affect each team's positions, work content, and hiring plans.
The controversy over AI entering the game world has never stopped. Supporters say it can accelerate prototyping, coding, and asset creation, cutting repetitive labor; worriers fear over-reliance will harm creative distinctiveness and increase the burden of review, while also raising old issues of copyright ownership, training data sources, and employment stability. Epic previously tested an AI-driven Darth Vader character in Fortnite, which produced inappropriate dialogue, and the team only fixed it through subsequent updates—showing that even when AI is confined to specific scenarios, generated results can still go off track and require additional testing and error correction.
At present, this shooter was originally slated to launch in March 2027, but has been indefinitely delayed at the same time as its process overhaul—Epic is not casually adding AI to a project proceeding as planned, but trying to swap production models for a new solution while behind schedule. Neither Epic nor Disney has announced a new release schedule, nor clarified what proportion AI-generated content will ultimately account for. What can be confirmed is that the company is accelerating the infusion of AI into internal development, and this collaborative work assembling Marvel and Star Wars characters may become the first touchstone for testing the quality of the AI-first model: if it significantly improves efficiency while maintaining quality, major studios will redraw the division of labor between programmers, artists, and AI tools; if the cost of review and rework exceeds expectations, the efficiency dividend AI brings may be far less attractive than it looks on paper.