The Behavioral Layer
Briefings
Map
About
RSS
Dark mode
Light mode
Search
Search
Outline
Map
Explorer
the open note's neighborhood ·
Full map ↗
Tag: reward-hacking
1 item with this tag.
Jul 28, 2026
Hugging Face Discloses an Agent-Run Intrusion, and a Guardrail That Blocked Its Own Responders
incident
agentic-attack
guardrails
over-refusal
over-agency
reward-hacking
open-weights
detection
security