AIR just raised $50M to vet the skills and plug-ins AI agents pull in on the fly. That's not a scare story -- it's a receipt for a tradeoff every extensible agent has already made.
Read more →
GitHub's August 2026 Copilot release piles on session-management tools for reviewing AI-generated code — a sign of how much process it now takes to safely use a coding agent.
Read more →
New research shows most enterprises trust their AI agents aren't over-provisioned -- but only a third actually enforce least privilege. Here's why that gap exists.
Read more →
Tricentis just launched AI tools to grade whether AI-written code is safe to ship. It's a real fix for a real problem -- one that a visible, no-code workflow was never exposed to in the first place.
Read more →
OpenAI's own data shows 98% of its employees use Codex, but under 1% of individual ChatGPT Work subscribers do. That gap isn't about awareness -- it's about scope.
Read more →
A Fortune investigation shows even Google's payment protocol can't tell you who's liable when an AI agent ignores your instructions. Here's why that gap is structural, not a bug.
Read more →