Online Safety Monitoring for LLMs
Despite alignment training, LLMs remain prone to generating unsafe outputs at deployment time. Monitoring outputs online and raising an alarm when safety can no longer be assumed is therefore critical. We study a simple real-time monitor that turns a verifier signal from an external model into an alarm decision by thresholding, with the threshold calibrated via risk control. In experiments on mathematical reasoning and red teaming datasets, we show that this simple design is competitive with more advanced monitors based on sequential hypothesis testing.
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
- Linked via arxiv authorMona Schirmer →
Online Safety Monitoring for LLMs
- Linked via arxiv authorMetod Jazbec →
Online Safety Monitoring for LLMs
- Linked via arxiv authorAlexander Timans →
Online Safety Monitoring for LLMs
- Linked via arxiv authorChristian Naesseth →
Online Safety Monitoring for LLMs
- Linked via arxiv authorMaja Waldron →
Online Safety Monitoring for LLMs
- Linked via arxiv authorEric Nalisnick →
Online Safety Monitoring for LLMs