Microsoft is pulling back Claude Code licenses after costs spiraled out of control. Uber burned through its entire 2026 AI coding budget in four months. The productivity promise is real, but so is the bill.

Anthropic's Project Glasswing has used Claude Mythos Preview to find over ten thousand high- or critical-severity vulnerabilities across critical open-source infrastructure in its first month. The bottleneck has shifted from finding bugs to verifying, disclosing, and patching them fast enough.

Microsoft is canceling Claude Code licenses for its internal developers and redirecting them to GitHub Copilot CLI instead. The move signals how quickly enterprise AI tooling decisions can shift beneath teams building on third-party coding agents.

ModelRift ran a practical benchmark asking six AI coding tools to build the Pantheon in OpenSCAD. The results reveal which models can actually handle spatial geometry, not just basic syntax.

Superset is an open-source code editor built for running multiple AI coding agents simultaneously on your local machine. It has already attracted 11,000 stars on GitHub, signaling real developer interest.

Virgin Atlantic used Codex to ship a revamped mobile app on a fixed holiday travel deadline, reaching near-total unit test coverage and zero P1 defects. Here is what that means for teams building under real pressure.