GCD tightens jailbreak detection with acceptance and refusal anchors and guarantees safe outputs by pre-injecting refusal tokens, cutting false positives 52% versus GradSafe while adding minimal latency.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Gradient-Controlled Decoding: A Safety Guardrail for LLMs with Dual-Anchor Steering
GCD tightens jailbreak detection with acceptance and refusal anchors and guarantees safe outputs by pre-injecting refusal tokens, cutting false positives 52% versus GradSafe while adding minimal latency.