This GitHub Copilot weekly update, released September 4 and covering the August 31 cycle, expands AI model access, adds a new content security control, and rolls out several IDE level features across VS Code and JetBrains
What This GitHub Copilot Weekly Update Includes
Here is everything this GitHub Copilot weekly update actually changed.
According to GitHub’s official changelog, Copilot Pro+, Max, Business, and Enterprise subscribers now have access to Claude Fable 5.1. This adds to the existing model lineup available to higher-tier subscribers, giving teams another option when choosing which model handles a given coding task.
Gemini 3.8 Flash also rolled out this week, available across a wider range of tiers: Pro, Pro+, Max, Business, and Enterprise. Unlike the Claude Fable 5.1 rollout, this places the new model within reach of Copilot’s standard Pro subscribers as well, not just the higher paid tiers.
On the security side, the update introduces a specific control: the Copilot app and CLI now honor content exclusions across agentic workflows. Previously, content exclusion rules primarily governed single code completions. This extends that same protection to multi-step, autonomous coding sessions, where an AI agent might otherwise touch a wider range of files without the same exclusion boundaries applied.
The GitHub Copilot harness also reached general availability in the JetBrains plugin this week, following a period in preview. This is notable because it puts JetBrains users on more equal footing with VS Code users in terms of feature rollout speed going forward.
VS Code 1.136 shipped several changes of its own. Agent Merge entered public preview, a feature that automates pull request readiness by resolving feedback comments, failed checks, and merge conflicts without manual intervention at each step. Multi-root workspace support for Claude agent sessions became available as an experimental feature, extending agent sessions across multiple folder hierarchies in a single workspace. Chat backgrounds entered an experimental phase for visual personalization, and chat sessions now organize related conversations hierarchically with attention indicators showing which threads need a response.
What This Means for You:
This GitHub Copilot weekly update matters less for the headline features and more for one line buried in the security section.: content exclusions now apply across agentic workflows, not just single completions. For any agency running AI assisted development on client codebases, that is the detail worth tracking, since it directly affects how confidently proprietary client code can be kept out of an AI tool’s context during automated, multi-step coding sessions.
The JetBrains harness reaching general availability is the second detail worth noting. Development teams that standardized on JetBrains IDEs specifically to avoid being second-tier Copilot users can now expect feature parity going forward, which removes a real consideration from the IDE choice conversation.
This GitHub Copilot weekly update is exactly the kind of detail that shapes how safely AI tooling gets used on real client projects. Magnero’s custom website design team already works with AI assisted development tools as part of standard delivery, and updates like this one, tightening what code an AI agent can and cannot see, are exactly the kind of detail that shapes how safely that tooling gets used on real client projects.
