Last year, GitGuardian found 1.27 million AI-service credentials exposed on public GitHub — up 81% from the year before. The root cause is predictable: every team shipping an LLM-powered feature starts by dropping a static API key into a CI secret, a container environment variable, or a .env file that eventually ends up in a commit. Anthropic’s answer to this is now generally available. As of June 17, Workload Identity Federation (WIF) is live on the Claude Platform. Your workload presents an OIDC token from the identity provider you already run. Anthropic validates it and returns a short-lived access token. […]
The post
source & further reading
byteiota.com — original article
Stack Overflow for Agents: AI Coding Memory Layer Lands
AutoJack: One Web Page Can RCE the Host Running Your AI Agent
GLM-5.2 Hallucinates 3x Less Than GPT-5.5 — Open Weight Wins