Microsoft · Filed Feb 18, 2026 · Published Jul 2, 2026 · verified — real USPTO data

Microsoft Patents a System That Uses AI to Rewrite a Game's Story Around You

What if a video game's story reshaped itself based on every choice you made, so that no two players ever experienced quite the same arc? That's exactly what Microsoft is patenting.

Microsoft Patent: AI That Rewrites Games Based on How You Play — figure from US 2026/0183672 A1
FIG. 1A — rendered from the official USPTO publication PDF.
Publication number US 2026/0183672 A1
Applicant Microsoft Technology Licensing, LLC
Filing date Feb 18, 2026
Publication date Jul 2, 2026
Inventors Sudha RAO, William Brennan DOLAN, Christopher John BROCKETT, Weijia XU, Nebojsa JOJIC, Gabriel A. DESGARENNES, Yun Hui XU
CPC classification 463/23
Grant likelihood Medium
Examiner CENTRAL, DOCKET (Art Unit OPAP)
Status Docketed New Case - Ready for Examination (Mar 25, 2026)
Parent application is a Continuation of 18213868 (filed 2023-06-25)
Document 20 claims

What Microsoft's AI-driven game storytelling actually does

Imagine playing a story-driven game and noticing that it seems to know what you find interesting. You keep avoiding combat and talking your way through problems, and suddenly the game starts offering you more of those moments. That's the core idea here: the game watches how you play and uses AI to generate new story content that fits your patterns.

Most narrative games today are built on a fixed map of choices and outcomes. Microsoft's patent describes a system that can bend that map in real time, adding new story branches or adjusting existing ones based on what individual players actually do, not what designers assumed they would do.

The AI doing the rewriting comes from the same family of models as ChatGPT, so it can generate dialogue, plot beats, and events that feel coherent, not just random. Think of it as a storytelling system that treats your play style as ongoing feedback.

How the system tracks choices and rewrites the story map

The patent describes a system built around what it calls a narrative state graph, a structured map of all the possible events, choices, and outcomes in a game. Each point in the story is a node; the connections between them are the paths a player can take.

The system does three things simultaneously:

  • Tracks the player's path through the graph, logging which branches they take and which they skip.
  • Monitors interactions with game assets, meaning it watches how players engage with characters, objects, and story moments, measuring things like time spent, choices made, and actions taken.
  • Uses a generative model (such as a GPT-style large language model) to modify the graph itself based on that data, creating new nodes or adjusting existing ones to better match what each player seems to respond to.

The engagement tracking piece is important: the system isn't just reacting to explicit choices. It's also inferring what a player finds compelling based on behavioral signals, then using that to guide the AI's content generation. The result is a story structure that evolves around individual play patterns rather than presenting every player with the same fixed experience.

What this means for the future of narrative games

For players, this could mean story-driven games that feel genuinely personal rather than branching in the same two or three ways for everyone. A player who gravitates toward exploration might get a richer world to investigate; a player obsessed with a minor character might find that character's role expanded. The game adapts to your behavior, not a designer's prediction of it.

For Microsoft, this fits squarely into its broader AI strategy. The company owns both a major AI research arm and Xbox, one of the largest gaming platforms in the world. A patent like this suggests Microsoft is thinking about how to apply its generative AI investments directly inside games, not just in the tools developers use to build them.

Editorial take

This is one of the more genuinely interesting patents Microsoft has filed in the gaming space in years. Procedural generation in games isn't new, but applying a large language model to dynamically restructure a story graph based on real player behavior is a meaningful step beyond 'random dungeon layouts.' Whether it ships in a real game anytime soon is another question, but the technical ambition here is real.

Which company should we read for you?

We track 17 companies here. Pro is the same weekly breakdown for any company you choose, delivered privately. Type a name and we'll scope it and send you a quote.

Get one Big Tech patent every Sunday

Plain English, intelligent commentary, no hype. Free.

Source. Full patent text and figures from the official USPTO publication PDF.

Editorial commentary on a publicly published patent application. Not legal advice.