HELM Safety

Search everything

Mentions · 7d0
Mentions · 30d2
All mentions2
First seenOct 1, 2026
Last mentioned22d ago

Timeline

2 articles · newest first

Searching for "Harmful Refusal": A Psychometric Audit of an AI Safety Benchmark

A psychometric audit finds HarmBench and HELM Safety do not isolate a single harmful-refusal attribute.

arXiv cs.AI / cs.LG / cs.CLOct 8, 17:46 UTC · 22d agoAI safety & security2 · 1 read

False Floors: LLM Safety Routing Evaluations Break Under Distribution Shift

Safety-router benchmarks overstate routing gains by picking the best model on test data.

arXiv cs.CROct 1, 12:09 UTC · 29d agoAI safety & security1 · 1 read

Appears with

HarmBenchproduct · 1
GPT-5.4ai model · 1
AgentDojoproduct · 1

Entities are extracted by the model from each article. First seen Oct 1, 2026, last mentioned 22d ago. Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.