Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Gradient-Controlled Decoding: A Safety Guardrail for LLMs with Dual-Anchor Steering

About

Large language models (LLMs) remain susceptible to jailbreak and direct prompt-injection attacks, yet the strongest defensive filters frequently over-refuse benign queries and degrade user experience. Previous work on jailbreak and prompt injection detection such as GradSafe, detects unsafe prompts with a single "accept all" anchor token, but its threshold is brittle and it offers no deterministic guarantee that harmful content will not be emitted once decoding begins. We introduce Gradient-Controlled Decoding (GCD), a training-free guardrail that combines an acceptance anchor token ("Sure") and refusal anchor token ("Sorry") tightening the decision boundary and significantly lowering false positives. In the mitigation stage, if a prompt is flagged, GCD preset-injects one or two refusal tokens ("Sorry, I can't...") before autoregressive decoding resumes, guaranteeing first-token safety regardless of sampling strategy. On ToxicChat, XSTest-v2, and AdvBench, GCD reduces false positives by 52% vs. GradSafe at comparable recall, lowers attack success rate by up to 10% vs. the strongest decoding-only baseline, adds under 15-20 ms latency on an average on V100 instances, transfers to LLaMA-2-7B, Mixtral-8x7B, and Qwen-2-7B, and requires only 20 demonstration templates.

Purva Chiniya, Kevin Scaria, Sagar Chaturvedi• 2026

Related benchmarks

TaskDatasetResultRank
Safety ClassificationAdvBench
F1 Score99.9
10
Content Safety ClassificationXSTest
Precision92.38
7
Content Safety ClassificationToxicChat
Precision67.34
6
Showing 3 of 3 rows

Other info

Follow for update