AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

How Catastrophic is Your LLM? Certifying Risk in Conversation

Chengxiao Wang, Isha Chaudhary, Qian Hu, Weitong Ruan, Rahul Gupta, Gagandeep Singh

ICLR 2026 · Large language models › Language, safety and interpretability · approach: Safety alignment of LLMs · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
constrain final query to harmful target set (graph path) vs vanilla last query (ablation)graph path distribution · HarmBench cybercrime, GPT-4o judgecertified lower bound: 0.403 -> 0.533 (Mistral); 0.123 -> 0.195 (Claude)Improved
use adaptive-with-rejection sampling instead of random node with jailbreak on safer LLMs (ablation)gpt-oss / Claude · HarmBench chembio / cybercrimecertified lower bound: 0.028 -> 0.101 (gpt-oss chembio); 0.028 -> 0.195 (Claude cyber)Improved
add jailbreak prefixes to random node queries (ablation)Nova Premier / Claude · HarmBenchcertified lower boundNo clear effect
increase jailbreak insertion probability (ablation)DeepSeek / Mistral / Llama · HarmBenchcertified lower boundImproved
increase jailbreak probability (ablation)Claude / gpt-oss · HarmBenchcertified lower boundNo clear effect
increase query sequence length (ablation)Llama / DeepSeek / Claude · HarmBenchcertified lower boundImproved
increase query-set size from 50 to 150 (ablation)certification · HarmBench scenariocertified boundsNo clear effect
vary AwR weight ratio, graph thresholds (ablation)certification · HarmBench scenariocertified boundsNo clear effect
certify over distribution instead of fixed single-turn evaluationcertified lower bound vs ST ASR · HarmBench cybercrimenumber of specs exceeding baseline: 157 of specs (DeepSeek) vs ST; 22 vs MTImproved
use certified distributions vs fixed multi-turn baselinecertified lower bound vs MT ASR · HarmBench chembionumber of specs exceeding baseline: 30 (Mistral), 29 (DeepSeek), 1 (Claude)Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy