Epic Games Goes All-In on AI for Disney Shooter Project
A Bloomberg report published this week revealed that Epic Games has restructured its development pipeline for its high-profile collaboration with the Walt Disney Company around a single, sweeping directive: artificial intelligence writes the code, AI guides the art, and human engineers manage the output rather than produce it themselves. The report, citing people familiar with the project who asked not to be identified, also notes that Bloomberg reviewed internal employee messages that confirm the strategic pivot. Those messages used stark language — "no more hand-written code" was one phrase cited — leaving little ambiguity about what Epic's leadership expects from its developers going forward.
The project in question is a Marvel- and Star Wars-themed shooter that was supposed to hit the market in March 2027. That window is now gone. The title has been delayed indefinitely, a consequence of both the operational turbulence caused by this AI-first overhaul and development that was already moving slower than expected due to understaffing. The stakes are not abstract. Epic and Disney formalized their collaboration in early 2024, committing $1.5 billion to the partnership. That figure alone makes this one of the most expensive joint ventures in gaming history, and the industry is paying close attention to whether the AI-first bet will justify the scale of that investment.
What 'AI First' Actually Means for Game Development
The term "AI-assisted development" has been in circulation for several years, but what Epic is describing is categorically different from a developer using GitHub Copilot to autocomplete a function. According to the Bloomberg report, Anthropic's Claude — one of the leading large language model platforms currently deployed in enterprise software environments — is among the AI models being used to generate code for the project. Art assets, too, are being produced through what the internal messages describe as "full AI workflows."
Read next Laika's Wildwood: Stop-Motion Fantasy at TIFF 2026Human engineers are not being removed from the process entirely. Their role has been redefined: they review, manage, and guide the AI-generated output rather than writing from scratch. This is a meaningful distinction. In the current model, a developer's primary skill set shifts from writing code to evaluating code — understanding what the model produced, catching errors, directing the model toward better outputs, and integrating generated components into a coherent whole.
To understand how significant this shift is, consider baseline adoption data. Stack Overflow's 2024 Developer Survey found that roughly 62 percent of professional developers were already using or planning to use AI tools in their development process. A McKinsey analysis from the same period estimated that generative AI tools could automate between 20 and 45 percent of software engineering tasks, depending on the complexity of the work. Epic appears to be pushing well past those percentages, treating AI generation not as a productivity supplement but as the primary production method.
Why the Marvel and Star Wars Shooter Is Now Delayed Indefinitely
The delay announcement carries its own significance. A project backed by $1.5 billion and anchored to two of the most commercially powerful intellectual property ecosystems in entertainment — Marvel and Star Wars — does not slip indefinitely without consequences. Losing the March 2027 window means losing a planned release date that was presumably coordinated with Disney's broader IP strategy, which operates on tight calendars tied to theatrical releases, merchandise cycles, and platform exclusives.
The Bloomberg report attributes the delay to two overlapping problems. First, the project was already underperforming against its development schedule due to understaffing. Second, the AI-first restructuring introduced new operational complexity at a moment when the project could least afford disruption. Rebuilding workflows, retraining engineers to operate as AI supervisors, and validating AI-generated code and art against quality standards takes time — time that was not built into a production calendar set before this pivot was decided.
This sequence matters. The AI-first strategy may ultimately accelerate development once the transition stabilizes, but the near-term cost is a timeline casualty. For a project of this scale, "indefinitely delayed" is not a minor setback.
What This Signals for the Broader Game Industry
The broader game industry has been watching AI development tools with a combination of genuine curiosity and institutional caution. Most studios experimenting with AI have done so at the margins — using it for localization, quality assurance testing, procedural content generation, or writing assistance. A handful of smaller studios have built entire workflows around AI-generated assets. But a AAA studio committing to AI-written code as the default for a blockbuster title tied to Disney IP is a different order of magnitude.
What Epic is attempting is closer to what some technology analysts have been predicting for several years: that generative AI would eventually become the primary production layer in software development, with human expertise providing direction and judgment rather than manual labor. Epic's decision effectively treats that prediction as a present-day operating assumption rather than a future possibility.
The implications for labor are direct and uncomfortable. If AI can generate production-ready code for a complex, networked, graphically intensive shooter — and do so at a pace that justifies restructuring a billion-dollar project around it — then the demand profile for certain categories of engineering talent shifts materially. Studios that have been hiring large teams of systems programmers, graphics engineers, and gameplay coders are now watching a major industry player bet that a smaller team of AI supervisors can produce equivalent output.
Risks and Criticisms of an AI-Driven AAA Game Studio
The risks here are real and not hypothetical. Large language models, including Claude, generate plausible-looking code that can contain subtle logic errors, security vulnerabilities, and performance issues that human engineers who did not write the code may be slower to catch than engineers who did. Code review is a skill; reviewing AI-generated code at scale, across an entire production codebase, is a different skill — and it is not yet clear that the industry has enough engineers trained to do it well.
Art generation carries its own complications. AI image and asset generation tools have faced sustained criticism over training data provenance, and studios using full AI art workflows for a Disney-licensed product would presumably require Disney's legal and creative teams to sign off on those workflows explicitly. The report does not detail how that aspect of the collaboration is being managed.
Quality assurance also becomes harder, not easier, when the people running QA did not write the code being tested. Familiarity with a codebase — knowing why a particular function exists, what edge cases it was designed to handle, what assumptions it encodes — accelerates debugging. That institutional knowledge is thinner when the code was generated rather than written by an identifiable engineer.
What Happens Next: Industry Reactions and Implications
Epic's public communications on this topic have been limited to what Bloomberg's sources have already described. The company has not issued a formal statement confirming or detailing the AI-first pivot as of publication time. That silence is itself a signal — studios making controversial workforce and production decisions rarely amplify them before the decisions have produced demonstrable results.
What comes next will depend on execution. If Epic can ship a polished, commercially successful Disney-licensed shooter built on AI-generated code and art, the industry will take notice in ways that no amount of internal messaging could produce. That would validate the model in the most unambiguous terms possible: a high-budget, high-visibility title delivering on its promise.
If the project struggles — if the AI-generated codebase proves brittle, or the art feels generic, or the development timeline extends further than the current indefinite delay — the cautionary argument against AI-first development will have an equally prominent case study attached to it.
The Epic Games AI code development story is ultimately a test of whether the tooling has matured enough to carry AAA responsibility. The answer will arrive eventually, in the form of a game — or the absence of one.
Source: Ars Technica - All content
