Shared Copilot Automations are rolling out so teams can build and maintain AI-powered workflows together. They belong to the repository, are billed to its owner, and act as Copilot rather than as an individual.
Image generation and editing by natural language arrives in GitHub Copilot as a public preview, powered by GPT Image 2.5 Flare and billed in image tokens against AI credits.
The open source GitHub Next project that runs coding agents inside GitHub Actions from Markdown definitions reaches general availability, with ARC support, model routing, and stronger agent isolation.
At Universe, the GitHub Copilot app gained Projects for multi-repository work and computer use on macOS and Windows, plus connected-tool context and in-chat image generation.
The JetBrains plugin gets Copilot Fusion in place of Auto, generally available local sandboxes and next edit suggestions, a new default agent harness, and default session permissions.
Enterprises working with a GitHub or Microsoft account team can get Copilot included for every GitHub Enterprise user, with usage metered against the enterprise budget instead of per-seat credits.
GitHub describes a context and learning layer for agents built on its data graph: Copilot Memory is generally available, dreaming enters public preview, and Copilot can search code across repositories.
The GitHub security extension lets you browse code scanning alerts, ask Copilot to explain or fix them, and run /security-review without leaving the Copilot app.
Copilot Fusion routes and orchestrates models per task at three effort levels, and an experimental preview adds routing between on-device and cloud models.