All game AI news

Mine Odyssey proposes an agentic spatial intelligence benchmark with 180 tasks, rebuilding 30 real locations in Minecraft

Yuxuan Cao and others propose Mine Odyssey, an agentic spatial intelligence benchmark that rebuilds 30 real locations across 20 countries and regions on five continents in Minecraft, with 180 natural language tasks, 20 of them outdoor scenes and 10 indoor scenes.

MultiWorldBench: can independently controlled views describe one shared world

The authors propose MultiWorldBench, a diagnostic Minecraft benchmark for testing whether independently controlled views in a multi-player world model stay consistent with one single persistent shared world; it contains 495 case configurations, seven task suites and ten capabilities.

Can AI agents learn their way to the top? AAArena evaluates heuristic learning in a long-running game competition

The paper proposes the AAArena benchmark, using 12 real adversarial games and 1,920 archived human programs to evaluate agents' learning ability in long-running competitions. Among the evaluated model and tool configurations, Opus5.5 with Claude Code took 6 gold medals, and no configuration could take the remaining 6 human ladders.

AgentGarten: a code world framework for continually evolving agents

The paper proposes the AgentGarten framework, which couples simulators and game engines to a shared neural renderer to build real-time interactive virtual environments. The simulation backend maintains persistent world state and executes program-defined interaction rules, while the renderer generates visual observations from structured conditions exported through a unified interface; the neural renderer is adapted from a pretrained video model to take geometry-conditioned input, is distilled with the proposed Adversarial Forcing, and has its inference optimized for real-time interaction.

miHoYo's Dawei becomes the biggest "human easter egg" at Xiaohongshu RED LAND as players say he never turned down a photo

miHoYo president Dawei appeared at Xiaohongshu RED LAND to back the company's new game Xingbu Gudi, and was called by players the biggest "human easter egg" of this year's event, posing for photos with everyone who asked and even appearing on a parade float. This year's RED LAND expanded from 80,000 square meters to 150,000 square meters, with more than 160 game and anime IPs taking part and five rounds of ticket sales selling out in an average of 20.6 seconds. For the first time, miHoYo's Genshin Impact, Honkai: Star Rail, Zenless Zone Zero and Xingbu Gudi appeared together offline, and Dawei also met the physical robot of Nalo, the AI café proprietress from Xingbu Gudi.

Google and Unity partner to launch AI game development platforms Playground and Unity Spark

Google and Unity announced a strategic partnership to jointly launch an AI game development platform for next-generation interactive entertainment, with Google's experimental platform Playground going live the same day and Unity Spark to be integrated later this year.

Minion Masters team BetaDwarf's new game Dragon Lair Breakers demo hits Steam, with Square Enix as global publisher

The demo for Dragon Lair Breakers, the multiplayer extraction ARPG from Minion Masters developer BetaDwarf, is now live on Steam, and Square Enix has announced it will serve as the game's global publisher. The demo offers 2 Breakers and 2 PvE difficulties, playable solo or with friends on the first chapter's map. The PvPvE test starting October 15 will add more Breakers and player-versus-player combat, with matches won by completing an extraction.

GDC: How Ghost of Yotei teaches faction behavior early in the game

In a #GDCRewind video, GDC breaks down how the Ghost of Yotei team taught faction behavior during the game's opening phase, with the full session available in the GDC Vault. The talk is titled Meaningful Reasons to Wander: Creating Open World Engagement in 'Ghost of Yōtei' and focuses on the design of motivation for open world exploration.

Epic launches Unreal MCP for Unreal Engine with official support for agentic LLM tools in the editor

Epic launched the Unreal Model Context Protocol (MCP) for Unreal Engine, officially supporting the connection of commonly used agentic LLM tools directly into Unreal Engine. The protocol is built on the open MCP standard, letting agents perform operations in projects through the capabilities exposed by each editor. Developers can use compatible agentic coding tools, or build integrations, skills, and workflows around their own way of creating; documentation is at https://epic.gm/ue-mcp-docs.

Unreal Engine shares tutorials for connecting MCP in UE and UEFN, plus a PCG workflow

Unreal Engine has put out tutorial videos on connecting MCP in UE and UEFN, along with workflow examples for PCG and MCP. Two YouTube videos demonstrate how to connect MCP in UE and UEFN respectively, and a Fab link points to UE City Sample for exploring how PCG and MCP work together.

A First Look at Unreal MCP in Unreal Engine 5.8

Epic has introduced Unreal MCP (Unreal Model Context Protocol) in Unreal Engine 5.8, giving agentic LLM tools an officially supported way to connect.

Same story as the pick “Epic launches Unreal MCP for Unreal Engine with official support for agentic LLM tools in the editor”

Glen Schofield on His Retirement and The Callisto Protocol's Development, Says a Good Idea Saves More Money Than AI

Veteran developer Glen Schofield, now retired, looks back on the development of The Callisto Protocol in an interview and says a good creative idea saves more money than AI can right now. He says Krafton decided to move the game's launch up to December 2022, earlier than the March 2023 he had been told, so planned content was cut back, and he later shipped 86 patches and all of the DLC.

Google Releases Its Own Game Creation Platform

Google has released Playground, an experimental game creation platform where users can generate and playtest games in the browser through conversation, without writing any code. Google Labs says the platform offers a blank canvas, starter prompt templates and guided support, so users can adjust physics, rules, characters and environments at any time, then share what they build with friends or the Playground community. More than a dozen games are already on the site, shown by player rating and play activity, and all of them have to pass a safety review.

Same story as the pick “Google launches experimental AI game platform Google Playground”

Chongqing's game industry faces a turning point as mini-games and PC and console teams may form two separate ecosystems

Chuapp visited Chongqing teams including Pathea Games and Youthcat Studio, laying out the turning point and bottlenecks the local game industry has hit after producing successful titles. Pathea Games currently has more than 300 employees, and when Dyson Sphere Program launched in 2021 Youthcat Studio had only 5 people in development, relying on outside teams in Shanghai and elsewhere for testing, community and marketing. The Chongqing mini-game industry base unveiled in October 2025 has brought in 34 companies with more than 200 office staff, but PC and console teams with longer development cycles are still constrained by talent and long-term funding.

Unity Launches Unity Spark, a Browser-Based AI-Assisted Game Creation Platform

Unity has launched Unity Spark, a browser-based game creation platform where users can describe, build, modify and publish interactive experiences in natural language, without having to learn to code first. The platform plugs directly into the Unity Asset Store, using thousands of 2D and 3D assets made by community artists, and according to the Financial Times it can also generate some assets, though most images and effects come from the existing asset library, with contributing artists receiving credit and payment.

Same story as the pick “Why Unity built the browser game creation tool Unity Spark”

Digital Puppets shares its stylized character workflow with Character Creator 5 and Headshot 3

Digital Puppets co-founder Antony Evans demonstrated how to use the AI plugin Headshot 3 together with Reallusion's Character Creator 5 and iClone 8 to turn AI-generated concept art into rigged, animation-ready stylized 3D characters.

Inworld breaks down TTS-2 voice steering and how to measure it

Inworld AI Head of Research Aleksey Tikhonov explains how voice steering in TTS-2 works: using instructions to tell a speech model how to deliver a line. The post focuses on how to measure whether those instructions actually take effect — automated audio judging fell short in its tests, so the team turned to measuring changes in the speech itself. He also discusses why instruction wording matters and why the same direction produces different results across voices; these metrics are used to track changes between model versions, and they have their limits.

Arena Breakout S19 adds the card mode Arena Breakout: Blockade Line and launches Commander Mode in a Girls' Frontline 2 crossover

Since Season 19 went live on September 2, Arena Breakout has rolled out Hardcore Mode 2.0, Commander Mode and the card mode Arena Breakout: Blockade Line, and on September 30 it launched a crossover with Girls' Frontline 2. The card mode is turn-based deck building, with cards split into four types — commander, soldier, weapon and order — three fronts on the field, and 5 preset starter decks. Commander Mode lets players command a squad of AI soldiers with a built-in wheel or voice, and the crossover characters OTs-14 and Klukai can be added to the squad.

Wizards of the Coast Union Expands, Demands Company Stop Pushing Generative AI

The Dungeons & Dragons team at Wizards of the Coast and Team X, which handles Magic: The Gathering and Duel Masters, have joined the UWOTC-CWA union and are demanding that Hasbro and Wizards of the Coast leadership stop pushing generative AI. The company must decide by October 13 whether to voluntarily recognize the expanded bargaining unit.

Why Capcom's REX Engine AI Plans Were Misread as AI Game Generation

Last week Capcom introduced its plan to integrate AI tools into REX, its next-generation RE Engine, but several outlets read it as an engine for generating games with AI.

Unity Unveils Unity Spark, a Prompt-Based Game-Building Tool for Google's AI Game Platform Playground

Unity has entered a strategic partnership with Google, launching the prompt-based game-building tool Unity Spark inside the developer environment of Google's AI game platform Playground.

Same story as the pick “Why Unity built the browser game creation tool Unity Spark”

Unity launches Unity Spark for no-code game creation and publishing on Google Playground

Unity launches Unity Spark, which lets users create games, iterate with others, publish and play without writing a single line of code; assets come from artist-made items in the Unity Asset Store and can be refined further in the new web editor.

Same story as the pick “Why Unity built the browser game creation tool Unity Spark”

Unity unveils Unity Spark, a web-based AI game creation platform

Unity unveiled Unity Spark, an AI game creation platform built on the Unity engine. It enters beta later this year and runs entirely in the browser with no download required. Appearing the same day as Unity Spark was Playground, Google's platform for generating games from natural-language prompts.

Same story as the pick “Why Unity built the browser game creation tool Unity Spark”

Kuration SDK: Addressing the Virtual2Real Gap in World Models via Data Curation

The paper proposes and open-sources along with it Kuration SDK, a general-purpose physical AI data curation toolkit that uses data curation to address the Virtual2Real gap in training action-conditioned world models. The authors train and evaluate diffusion world models on CounterStrike gameplay data, confirming that metrics such as FVD, LPIPS and JEDi do not correspond to qualitative playability, and argue that curating raw gameplay data before training begins and measuring multiple diagnostic properties is a more reliable signal.

LeCuration: A Tiny World Model as a Data Curation Multi-Tool

LeCuration is a tiny world model used to curate data for another, larger downstream model. It uses LeWorldModel (LeWM) as a latent encoder and predictor, and adds a DiT decoder to supply visuals for autoregressive gameplay rollouts; its embeddings can serve as an anomaly detection signal and a content-based clustering heuristic, while autoregressive prediction of game state can be used to qualitatively check whether actions and states are consistent. The team ran a qualitative proof of concept on CS:GO gameplay data and has not yet reported quantitative curation metrics or downstream training results.

Why Unity built the browser game creation tool Unity Spark

Unity partnered with Google to launch Unity Spark, which lets users make and share games directly in the browser without Unity or game development experience. It connects to the Unity Asset Store with access to thousands of community-made art assets, and targets players, graphic designers and modders who are familiar with a genre and spot the gaps in it but don't know how to code.

Google launches experimental AI game platform Google Playground

Google launched an experimental AI game platform called Google Playground, where users can create, play and share fully customized games by entering prompts through a conversational interface, with no coding experience required. Google says users can direct it like a director, adjusting rules with prompts, customizing characters and fine-tuning environments, and the browser-based platform runs games on phones and laptops. Unity announced Unity Spark, a new tool that goes with Playground, at around the same time.

Muse Spark 1.3 goes live on Inworld Realtime Router

Inworld AI announced that Meta's reasoning and coding model Muse Spark 1.3 is now live on Inworld Realtime Router, with closed weights, support for a 1M token context, and pricing of $1.25 per 1M input tokens and $4.25 per 1M output tokens. According to its disclosure, in Meta's internal coding tests, this version uses about 25% fewer tokens than Spark 1.2.

Paper proposes Recursive Game Creator, using four recursively iterating components to improve generated game experience

The paper proposes Recursive Game Creator, an experience-oriented agentic game development framework that uses four components — Designer, Builder, Player and Reviewer — to iterate recursively and push a rough game prototype toward a more replayable work.

A space-agnostic visual game analytics tool for HoloLens 2: adaptive spatial reconstruction and realtime object detection

A space-agnostic visual game analytics tool for mixed reality (MR) game development has been proposed, combining adaptive spatial reconstruction with realtime object detection and running on the Microsoft HoloLens 2 MR headset. The tool is designed for the complexity of MR games, which must integrate real-world and digital components, and for the difficulty existing visual analytics tools have in achieving space-agnostic generalization. The researchers hope it will deepen understanding of players' surroundings and open a new direction for visual analytics in MR game development.

Study proposes ResNet-BiLSTM for multi-label perceptual bug detection from gameplay footage

The study proposes a ResNet-BiLSTM model that performs multi-label perceptual bug detection on gameplay footage, achieving an F1 score of 85.78% on a benchmark dataset, and compares it with video classification models including Inflated 3D ConvNet and 3D ResNet.

Sega Says It Won't Entrust the Creative Aspect of Games to AI, and Is Using the Tech to Improve Efficiency in Publishing and Other Departments

Sega senior executive director and deputy chief commercial officer Tsuyoshi Saito said the company is very cautious about using generative AI in game development and will not entrust the creative aspect of games to AI; it is currently used mainly in publishing and corporate departments to improve efficiency and generate business ideas.

MistralAI tweet

<translation> ``` If this tweet itself is also a pure link with no text, then please also say so—that means there is no body text to translate, and no title or summary needs to be generated.

SpeedrunBench launches to benchmark frontier LLM agents' strategy formation with speedruns across 9 games

The paper introduces SPEEDRUNBENCH, a speedrunning benchmark that evaluates frontier LLM agents on 9 different games. Agents must repeatedly refine their strategies over long-horizon actions, review their own performance and use the knowledge they have gained to be faster than their past selves and others. Experiments show frontier agents approach human world records on simple platformers, but on longer, more complex games they still lag behind human performance under practical budget constraints; the authors see the benchmark as a long-term testbed for studying agents' strategy formation ability.

Attacca: Goal-Directed Control under State Continuity for Long-Horizon Embodied Agents

The researchers propose Attacca, which trains a vision goal-conditioned policy on complete search-to-interaction trajectories so long-horizon embodied agents can carry on to the next task from the position, orientation and world state left by the previous one. The method uses context-decoupled goal sampling, pairing each demonstration with a class-compatible masked goal image from another world, and provides supervision beyond action imitation through a goal mask prediction head, while introducing behavior phase conditioning that distinguishes the Search, Approach and Interact phases.

SENSE: a state-aware emotion navigation storytelling engine

The paper proposes SENSE, a state-aware framework for generating playable branching visual novels with multi-track affective navigation. The framework integrates a stateful narrative architecture called MIND, a structural analyzer and a path-aware context management module, and can generate multiple intersecting routes from a small number of high-level inputs while maintaining character consistency and narrative causality. Evaluations using LLM judges, affective metrics and visual evaluation show SENSE outperforms baselines in narrative diversity and asset integration robustness, and initial human playtests show a directional improvement in affective fidelity with comparable fun.

Emoception proposes the SALFT framework, selectively fine-tuning a video ViT to recognize changes in player arousal

The paper proposes SALFT (Selective Affective Layer Fine-Tuning), a framework that fine-tunes a Video Vision Transformer using a selection criterion based on layer-wise parameter L2 norm change to recognize changes in player arousal from gameplay footage.

Unity shows an agent turning concept art into a runnable UI Toolkit menu with the Unity CLI

Unity showed an agent workflow in which handing concept art and a short description to an agent is enough to build a runnable UI Toolkit main menu through the Unity CLI. The agent also runs Play mode UI tests itself and captures Game view screenshots to check its changes and keep iterating, with the user guiding the process. The original includes a link to the full tutorial video.

PlaySuite: a large-scale benchmark for interactive visual intelligence covering more than 5K open-source games

The authors propose PlaySuite, a large-scale benchmark for evaluating interactive visual intelligence built on more than 5K open-source games from PyWeek and itch.io, spanning engines including Pygame, HTML5, Godot and Unity.

Unity demos prototyping a UI Toolkit main menu from concept art with the CLI and an AI agent

Unity has released a tutorial video showing how an AI agent, working through the Unity CLI and starting from concept art and a short brief, can build a usable main menu interface directly inside the Unity editor.