I'm excited to share Erik's work! "Mastering Atari 2600 Games with Discovered Options": we propose a general option discovery method that, under high-dimensional observations and tabula rasa conditions, learns from a single stream of experience and outperforms DreamerV3, Rainbow and IQN! 1/9 Looks very promising!
Three groups of Eggy Party UGC creators describe how they went from making maps as a hobby to forming teams and turning park maps into long-running products. The Gemini Editor announced at the official July creator conference runs the frame-synced Origin Editor and the state-synced World Editor in parallel, and the September update added map preloading, asset hall cleanup, and the Tripo P2.0 model for AI model generation.
Why it matters: The three creators' accounts trace the Eggy Party UGC path from making maps to making products, offering a look at how platform tools and incentives support long-term commercial operations.
The paper proposes MATE (Mutual Attention for zero-shot TEaming), a multi-agent reinforcement learning method inspired by human joint attention that lets agents align visual attention on salient objects in the scene during interaction, so they can coordinate zero-shot with unfamiliar partners.
Meshy announced it is sponsoring the Unity AI Game Jam at G-STAR 2026 alongside OpenAI, a 24-hour event for 50 developers held on November 18–19 at BEXCO in Busan. According to the organizers' breakdown, ideas use OpenAI, 3D uses Meshy, and development happens in Unity, with Unity and K-GAMES as hosts. Registration is first come, first served and may close early if applications are high; event details are at on.unitysquare.co.kr/4xRg3D8.
Wei Wu proposes Code2Games, an agent framework that lets coding agents build structured game worlds on top of a Blender world generated from the same game intent. The framework coordinates scene analysis, gameplay planning, constrained game world generation and engine customization through shared scene and gameplay representations, and when adapting to Unreal Engine 5 uses compilation diagnostics, runtime feedback and playtest results for execution-guided refactoring.
Why it matters: The paper coordinates the generation stages with shared scene and gameplay representations and compares generation quality on a benchmark containing ten fixed game prompts, making it easy to measure against existing approaches.
Warhorse Studios co-founder Martin Klíma gave an interview on the studio's 15th anniversary, looking back on a career that ran from translating Dungeons & Dragons to developing Kingdom Come: Deliverance II. He argues that AI cannot bring game production costs down: even with a fivefold gain in efficiency, studios would only use the same budget to build a larger game. He also revealed that Warhorse has obtained a The Lord of the Rings IP license to develop a spin-off game, and has split the team into two to push forward two projects.
Why it matters: The founder's 200-person team and fivefold AI assumption explain why AI cannot push game production costs down.
GPT-6.1 Sol is now live on Inworld Realtime Router. - OpenAI's upgraded Sol tier - According to OpenAI, intelligence approaching Astra at one-fifth the price - 1M token context, 128K output - $2.00 per 1M input tokens / $10.00 per 1M output tokens. At provider rates, no markup. Just swap the model string and use it directly!
GDC released a replay clip discussing how the Kingdom Come: Deliverance II team handles crime perception at far and mid LOD. The material comes from the session Supporting Thousands of NPCs in 'Kingdom Come: Deliverance' & 'Kingdom Come: Deliverance II'; the full session is available on GDC Vault: https://gdcvault.com/
Unity releases an official Grok Build plugin, making Grok Build the third coding agent to receive Unity engineering guidance.
Why it matters: The plugin shares the same skill repository as Claude Code and Codex, so readers can judge whether Unity guidance stays consistent when multiple agents develop in parallel.
An industry employment survey covering 90 countries and more than 1,800 professionals found that 35% of game company founders have cut hiring spend because AI tools can do the work. The BGIES report says 78% of respondents said AI changed the way they work over the past year, with 33% calling the change significant; companies increasingly split incremental work among existing staff, AI and outsourcing, and full-time work is becoming a privilege.
Why it matters: The report uses three data lines — hiring thresholds, outsourcing spend and median salaries — to show that AI's squeeze on game jobs starts before hiring.
Sanguo: Zhance Changhe, a Three Kingdoms SLG developed by Kaibo and published by Ruyi Jingxiu, is in a non-paying, data-wipe test on TapTap. Producer Aqing says the game focuses on visualized combat, reduced grind and screenplay-based short seasons, and that the mobile and PC versions are still awaiting license approval.
This is a big step in our work on speech-to-speech experiences, bringing speech understanding, reasoning and expressive voice generation closer together — and we are just getting started. The full article covers why we came together, what you can get today, and what's next: https://tinyurl.com/inworldultravox
Two game AI patents Microsoft filed with the USPTO this year are broken down point by point by AI and Games author Tommy Thompson, who finds little to look forward to in either.
Why it matters: The author compares these two Microsoft patents with earlier game AI patents from Sony and Nvidia, and goes through each point explaining why he is bearish on their prospects of shipping.
During the Mid-Autumn Festival, Zaomeng Ciyuan's AI streamer Ziying took on a time-limited broadcast challenge in her live room; the real-time interaction model behind her is Vidu S2, released not long before, and the room drew as many as 20,000 viewers at one point. Ziying has no fixed script: viewers can request songs, vote on her look and send gifts, and she has to change outfits and perform while interacting, then pick up where she left off after being interrupted. The author also tested Vidu S2's Avatar and Editing: uploading a real person's photo or an anime image and adding some settings creates a digital human, and switching on the camera allows real-time outfit changes, art style changes and background swaps while filming.
Why it matters: Using a livestream hosted by an AI streamer with no fixed script as a case study, it tests Vidu S2's real-time interaction and outfit-switching abilities, and discusses how they could be applied to game NPCs.
Tencent is quietly running a closed beta of E Ciyuan, an AI game companion product that offers digital human characters, voice chat and real-time screen recognition. Players can pick male or female companions with different personas for everyday conversation, or turn on a game companion mode. The product is free during the beta, and the interface already shows a framework for exchanging star coins for energy and for spending energy by conversation turn, voice or length of companionship; the launch date, the list of supported titles and the final pricing plan have not been made public.
Why it matters: The article places E Ciyuan alongside in-game companion examples such as Lingbao and AI Pom-Pom, showing the difficulties standalone AI companions face in adaptation and monetization.
Inworld has announced the acquisition of Ultravox, a realtime voice agent platform; the Ultravox team has joined Inworld and continues to develop the platform. The first update after the acquisition is that the built-in Inworld voice in Ultravox now uses Realtime TTS-2, existing voice IDs remain usable with no code changes, and the upgrade costs nothing extra.
Why it matters: The acquisition brings voice understanding, reasoning and voice generation into one realtime inference infrastructure, letting readers judge how the voice agent stack is consolidating.
Episode 6 of the AI and Games podcast brings in Andrei Muratov, one of the leads for AWS's games business, to talk about the realities of putting AI to work in the games industry. The two discuss how studios assess whether AI is viable, how to keep up with AI's expansion, and how AWS positions itself to support studios actually adopting AI. The podcast is presented by AWS for Games and was recorded on the Wednesday morning of Gamescom week.
On a livestream, AI and Games plays SPRAWL Zero and breaks down the game's NPC AI as it goes. SPRAWL Zero is a fast-paced shooter inspired by the golden age of 2000s FPS games, set in a cyberpunk city collapsing under military rule and cult fervor, where players fight with Gravity-Gloves, enhancement abilities and more than 40 weapons.
Rather than porting the previous game, RUST LTD rewrote H3VR2 from scratch in a brand-new Unity project, spending nearly six months first on architecture and tools. The team numbers more than 20 people, firearm internals are simulated by an in-house system, and ballistics are built on ECS for Unity and Burst; the project moved from H3VR to Unity 6.3, with an emphasis on the SRP Batcher, shader compilation improvements and tile post-processing.
AI and Games Conference 2026 has announced its first batch of talks; the event will be held on 10-11 November at Plexal in London.
Why it matters: The agenda lays out what teams such as Embark, Wargaming and Warhorse will cover, and the lineup shows how major studios are applying AI to animation, navigation and NPCs.
Free for two weeks! Hy Image3.5 preview is now live on OnSolo. Short drama character sheets. Full-motion video game assets. Keyframes. Characters stay consistent across every episode. Editing is refinement, not starting over.
NVIDIA announced this month's updates for RTX game developers, including DLSS 5 with 3D-Guided Neural Rendering controls, speech pipeline extensions and inference framework updates for NVIDIA ACE, and RTX Kit 2026.3, which includes RTX Mega Geometry 2.0. The official tweet links to the developer blog post for this update; capabilities and integration details are in the original on developer.nvidia.com.
Why it matters: The official post lines up DLSS 5, NVIDIA ACE and RTX Kit updates side by side, so developers can compare integration priorities and upgrade order for their own projects.
GDC's retrospective series released a clip from a talk on Keeper's flight system, in which the development team explains how they managed the game's memory budgets. The full session is titled Fixed Wing Flight in 'Keeper': Pathfinding, Locomotion, and Animation of a Bird Companion, covering pathfinding, locomotion and animation for a bird companion in fixed-wing flight; the full video is up on GDC Vault.
The AI Gaming Festival on Steam is hosted by US technology company Invirtu Labs and centers on showing how generative AI can take over development tasks such as art assets, code, narrative, localization, voice and music. The author notes that the event is connected to Glitch, the indie game platform owned by its organizer Devin Dixon, giving it the feel of promotion for his own platform. He also says the accompanying talks are in fact a paid virtual conference, and questions whether some of the participating games were added without their developers' knowledge.
Why it matters: The article traces the organizer's ties to its own platform Glitch and lays out why this Steam event is seen as drumming up support for generative AI rather than explaining it.
AI and Games examined how the AI in Command & Conquer and Red Alert is implemented, drawing on the source code EA released alongside Command & Conquer Remastered Collection and made available on GitHub.
Why it matters: Using the remastered edition source code EA made public, the article breaks down unit tasks, commander states and pathfinding in the two early RTS games.
Inworld AI has released Realtime TTS-2, a new-generation realtime conversational voice model, now fully available in the Inworld API and the Inworld Realtime API.
Why it matters: The original post covers multi-turn audio context, natural-language voice instructions and a first-audio latency of under 200ms, which readers can use to judge whether realtime voice conversation fits their own projects.
Inworld AI has open-sourced the TTS Open Evaluation Toolkit, a customizable TTS evaluation framework for sampling TTS systems, evaluating audio offline and generating comparable reports.
Why it matters: The post explains why the same TTS model can produce different WER in different teams' hands, and covers how to run this evaluation tool and its first set of metrics.
NVIDIA announced at Gamescom that publishers including EA, Embark and Ubisoft will bring games and anticheat support to RTX Spark, which launches this fall, with EA Javelin Anticheat also natively supporting the platform.
Why it matters: The partner list ahead of RTX Spark's fall launch grows from May's COMPUTEX to this Gamescom, so readers can compare the two batches of publishers to gauge how adoption of the new platform is progressing.
Forza Motorsport's AI was rebuilt from the ground up, and the Drivatar system no longer relies on human play data. In a video, AI and Games argues that what the rebuilt Drivatar shows players is not how it actually works.
Why it matters: The video explains why Forza Motorsport's ground-up rebuilt Drivatar no longer depends on human play data, and points out the gap between its behavior and what players see.
The Drivatar bot Bowie Knife99 in Forza has become a new rival the player community can't stop talking about, known for its aggressive driving style: it rams cars on the road and repeatedly pulls off extreme corner cuts. It is currently the most-discussed AI in the community.
In its 85th video, AI and Games explains why Drivatar "Bowie Knife99" in Forza Horizon 6 became both a community meme and a public enemy in online races, tying the phenomenon to Turn 10's changes to AI driver behavior in 2023's Forza Motorsport.
Why it matters: It traces the link between Bowie Knife99's rise to fame and Turn 10's changes to driver AI behavior in 2023's Forza Motorsport, offering a comparison with the older Drivatar system.
Inworld introduces @mastra/voice-inworld-realtime, which merges STT, LLM and realtime TTS into a single WebSocket session bound to a Mastra Agent.
Why it matters: The post contrasts cascaded voice pipelines with a single WebSocket session and provides a reference CLI in under 100 lines of TypeScript, which readers can use to judge the integration cost.
Inworld announced price cuts across its entire voice stack, with most developers seeing reductions of more than half at each layer of text-to-speech, speech recognition, LLMs and compute. The new prices are already in effect for all developers. Realtime TTS-2 drops to about $10/1M characters, speech recognition to about $0.10/hour, and Router can connect to 100+ models with no markup on third-party models.
Why it matters: Inworld has lowered prices at every layer of its voice stack and published tiered per-unit rates for TTS, speech recognition and LLMs, which can be used to estimate inference costs for consumer-grade applications.
Inworld teamed up with Stream to build a reference implementation, Crashout Buddy, combining the latest speech model Realtime TTS-2 with Stream's open-source Vision Agents framework, so it can see facial expressions, hear speech and adjust its delivery in real time.
Why it matters: The post shows how to turn visual signals and emotional context into speech delivery instructions, which readers can use to judge whether this realtime voice pipeline can be plugged into their own projects.