FlexGuard outputs calibrated continuous risk scores for LLM content and uses thresholding to adapt to varying strictness levels, showing better accuracy and robustness than fixed binary moderators on the new FlexBench benchmark.
**Only the text after`[Assistant]:`is assessed for risk.** The text after`[User]:`is **never assessed as risky content**
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
FlexGuard: Continuous Risk Scoring for Strictness-Adaptive LLM Content Moderation
FlexGuard outputs calibrated continuous risk scores for LLM content and uses thresholding to adapt to varying strictness levels, showing better accuracy and robustness than fixed binary moderators on the new FlexBench benchmark.