#security

Articles tagged with #security

Fable and ChatGPT Authorization Refusals: Designing AI-Assisted Access Control That Fails Safely
AI coding agents can turn a permission model into policies, tests, integration code, and documentation. That makes authorization a natural target for AI-assisted development. It also creates a reliabi
Aug 2, 20266 min read
Your AI Agent Should Not Be Allowed to Approve Its Own Tool Calls
Tool calls turn language into side effects. Sensitive actions need authorization outside the model.
Jun 4, 202613 min read4