Every published Top AI Stories item tagged with GitHub Copilot, newest first.
GitHub shipped a research preview that builds an execution plan at runtime and picks models from several providers to draft, critique, revise, or escalate to something stronger. On the TerminalBench 2.1 suite GitHub reports it scoring 4.9 points higher than Claude Opus 5 at 67 percent lower estimated cost, and landing within about a point on two other suites at 36 and 65 percent lower cost. Every figure is GitHub's own, from controlled offline runs. It is live on all Copilot plans through the experimental command in the Copilot command-line tool, with the app and the editor extension due this month.
Wiz Research says its autonomous Red Agent, running without human direction, found a command-injection flaw in a Snowflake GitHub Actions workflow that let anyone run code by opening an issue with a crafted title. The vulnerable change had shipped in a pull request co-authored by GitHub Copilot Autofix, which reviewed the same code and marked it clear. An exposed Jira token gave read access to Snowflake's engineering, security compliance and bug bounty projects. Snowflake patched the same day and says no outside party got in.
GitHub started rolling out Moonshot AI's Kimi K2.7 Code — an open-weight model from the Chinese lab — inside GitHub Copilot for Pro, Pro+, and Max subscribers, with business and enterprise plans to follow. GitHub hosts the model on Microsoft Azure and bills it by usage. It is a notable sign of Chinese open models reaching the West's most widely used coding assistant, though enterprise admins must switch it on manually before their teams can pick it.
GitHub switched Copilot from flat-rate subscriptions to usage-based, token-metered billing, and developers are furious. Power users running agentic coding sessions report their costs could jump as much as 10 to 50 times, with some $29-a-month plans projected to balloon past $700; GitHub also removed the unlimited fallback model. The change makes Copilot's real cost depend on how heavily each developer leans on AI, and rivals are already courting the backlash.
Microsoft also used Build to launch the GitHub Copilot app, a desktop technical preview that GitHub calls "a control center for agent-native development." It lets developers run multiple AI coding agents in parallel — each in its own isolated git worktree — and consolidates issues, pull requests, and background automations into one "My Work" dashboard. A new Agent Merge feature shepherds a pull request through CI checks, reviews, and merge. The preview spans Windows, macOS, and Linux for Copilot Business, Enterprise, Pro, and Pro+ users.
The Verge reports that Microsoft will stop issuing Claude Code licenses to its own employees effective June 30, redirecting users to GitHub Copilot CLI. The original internal pilot — meant to expose project managers and designers to AI coding for the first time — reportedly burned through Microsoft's 2026 AI budget in months, with one employee reporting Claude consumed their monthly token allocation in just over a week. Hacker News commenters framed the move as procurement economics, not contractual or security friction, and noted developers picked Claude over Copilot when offered the choice — undermining Microsoft's own product strategy.