guardrails AI News & Updates

Cybersecurity Community Criticizes Overly Restrictive Guardrails on Anthropic's Fable

June 10, 2026 15:41

Cybersecurity researchers are criticizing the safety guardrails on Anthropic's newly released Fable model, claiming it overly blocks benign inquiries related to coding and security. When triggered by safety keywords, Fab...

Anthropic Releases Fable 5 with Robust Guardrails and Recursive Self-Improvement Warnings

June 9, 2026 17:00

Anthropic has released Claude Fable 5, a publicly available version of its highly capable Mythos model designed for advanced reasoning, software engineering, and vision tasks. To mitigate safety risks, the model is equip...