Unity Technologies has released Unity 6.2, a mid-cycle update in the Unity 6 series, delivering a suite of integrated Unity AI tools directly within the Editor. This replaces the existing Muse and Sentis services with new Editor-native functions including Assistant, Generators, and the rebranded Inference Engine. Unity emphasises better integration, a broader selection of AI models, and updated pricing, although detailed pricing mechanics remain unconfirmed. Muse is now officially retired.

Unity 6.2 also introduces other features relevant to production artists, such as an automatic Mesh LOD system, updates to Shader Graph, improvements in rendering pipelines, enhancements to the 2D system, and refinements to the UI Toolkit.

Generative AI for assets—with vendor variation

The Generators feature centralises sprite, texture, material, sound, and animation generation workflows. It supports both Unity’s own AI models and third-party models from Scenario, Inc. and Layer AI, Inc., trained on foundations like Stable Diffusion and FLUX. Unity retains its proprietary model for generating tileable textures.

Assistant: an AI-powered Editor copilot

Assistant replaces Muse Chat, delivering a “project-aware” experience inside the Editor. It can answer documentation queries, batch rename assets, automate tasks, and write or run C# code. Assistant supports:

• /ask: documentation and project queries without modifying files
• /run: automates repetitive tasks, directly applying Editor changes
• /code: generates or reviews C# code snippets interacting with the Unity API

Local inference—Inference Engine replaces Sentis

The Inference Engine, formerly Sentis, offers local model execution within the Editor or at runtime. It supports various model formats and is intended to support future ML-Agents compatibility. Unlike the cloud-based AI, this component ensures local inference without data leaving the machine.

Data handling and user responsibility

By default, Unity does not use user data to train AI models. Developers may opt into sharing prompts and metadata via the Unity Dashboard—but Unity insists runtime applications or user-created media (like meshes or audio) are not used to train either Unity’s own models or third-party models.

Partner Model providers receive anonymised input—prompts, references, and other data—for processing but not to train their models. Unity shifts the legal risk to users:

“Unity users are ultimately responsible for ensuring that their use of Unity AI complies with our acceptable use principles” and that generated assets do not infringe on third-party rights.

Additional Unity 6.2 improvements

Beyond AI, Unity 6.2 includes updates such as support for Android SDK 36, improvements in XR development, and new diagnostic tools for performance monitoring and crash reporting. The Mesh LOD and revised UI Toolkit are also emphasised in Unity’s release notes.

Production implications

Unity 6.2 is now officially supported as a non-LTS (update) release and receives the same level of support as the Unity 6.0 LTS until the next release arrives. It’s intended for new or mid-cycle productions seeking the latest tools.

Summary for production artists

Unity 6.2 delivers tightly integrated generative AI tools within the Editor, replacing legacy services and giving artists multiple options to create assets and code. The local Inference Engine assures offline capabilities. However, asset ownership and IP clearance remain the developers’ burden. The release also adds practical improvements to LOD workflows, diagnostics, XR, and UI tools

The post Unity 6.2 Welcomes AI, But Pace Caution—User Liability on Copyright first appeared on DIGITAL PRODUCTION and was written by Bela Beier.