Xbox Labs turns AI game port chaos into a creator bet

Microsoft is forming Xbox Labs to study decompiled and recompiled games. Xbox CEO Asha Sharma says success means more people playing and creating, signaling a shift toward AI-assisted co-creation.

What happened

Microsoft has created a new Xbox Labs group to experiment with decompiling and recompiling games, according to Xbox CEO Asha Sharma. The group is described as “a small group of game makers, engineers and scientists whose job is to turn emerging ideas into things our players and developers can experience.”

Sharma’s announcement follows a wave of retro gaming mods: gamers have ported classic titles to new devices or merged code from two different games into one experience. Microsoft has seen skateboarding modded into Call of Duty and Halo played in a web browser, and the company now says it is officially embracing this new AI-powered era of game modding.

In a memo, Sharma wrote that recent weeks brought rapid decompiling and rebuilding of games to run on hardware they were never designed to support. Windows PC builds have been translated to run on Apple Silicon or ported to run inside a browser through WebAssembly and WebGPU, with WebGPU now approaching 90% global browser support.

Key facts

• Microsoft is putting together Xbox Labs, a small group of game makers, engineers and scientists, to explore decompiled and recompiled game mods.
• Xbox CEO Asha Sharma says Xbox Labs will “build in the open, learn from what people do, and measure success by whether more people are playing and creating.”
• Sharma says people spend more than 72 billion hours a year playing Xbox games, roughly five times the estimated time spent writing code worldwide.
• Recent mods include Windows PC builds running on Apple Silicon, browser ports via WebAssembly and WebGPU, and big games remixed into the source code of another; WebGPU is approaching 90% global browser support.
• Microsoft did not respond in time for publication when asked about unauthorized IP use in decompiles and recompiles.

Our analysis

Xbox Labs marks a shift from platform policing of fan modifications to an explicit internal experiment. For years, Microsoft’s Xbox backward compatibility program let old games run on modern hardware under Microsoft’s own business model restrictions. Now modders are doing similar work outside those restrictions, and Microsoft is trying to learn from them rather than simply block them. That likely changes how publishers and platform owners view AI-assisted remixes: as a source of engagement and co-creation, not only a legal problem.

The creator angle is the clearest signal for social and content teams. Sharma frames success as “whether more people are playing and creating,” and argues that making games easier to access and allowing players to become co-creators are “unmistakably good outcomes.” This suggests Microsoft expects AI agents and modding tools to lower the barrier between audience and creator, similar to how coding became the first major workload for agents because they can write, run, and test software.

Still, the unresolved IP question matters. The bigger question is how Microsoft will respond when modders take its IP and make it freely available, and the company did not comment in time for publication. That ambiguity is likely to shape how quickly developers and brand partners will embrace the new tools.

What it means for operators

• Track Xbox Labs tooling and learnings, not just game announcements. If Microsoft shares decompiling or AI remix tools with developers, social and content teams can test similar community-remix loops in their own channels.
• Build co-creation metrics. Sharma’s stated success measure — more people playing and creating — is a useful KPI for community managers: count remixes, variants, and user-generated responses alongside reach and engagement.
• Set IP boundaries before AI remixes scale. Microsoft’s lack of response on unauthorized IP use shows that even a platform owner can lag behind modder behavior; operators should decide now which fan works are allowed, credited, or monetized.
• Prepare for agent-assisted creation. Sharma compares game properties to coding, where agents can write, run, and test software; consider documenting repeatable content processes so AI agents can later handle parts of production without losing brand voice.