A red-teaming study documented failures and safety gaps in model handling of self-harm prompts.
Contents
Summary
2025-MLP-03 is a A1 benchmark study row involving Multi-LLM red-team (6 models). A red-teaming study documented failures and safety gaps in model handling of self-harm prompts.
Jailbreaking LLMs in Suicide/Self-Harm Contexts. Original date string: July 1, 2025 (posting). Source summary: A red-teaming study documented failures and safety gaps in model handling of self-harm prompts.
Coding
2025-MLP-03 is coded as UNSAFE_OUTPUT with grade A1. The public row assigns JB-AC, GEN-M, DET-FN and treats the incident as evaluation or demonstration.
| Field | Value |
|---|---|
| Platform | Multi-LLM red-team (6 models) |
| Model version | multi-LLM evaluation set |
| User type | test |
| Jurisdiction | unknown |
| Evidence type | benchmark_study |
| Product mechanism | benchmark_or_evaluation_set |
| Trajectory strict | no |
Assigned taxonomy codes
Evidence
The public evidence anchors for 2025-MLP-03 are the materials listed below. Only explicitly linked evidence files are hosted live.
Related incidents
The closest related pages are the incidents that share a provider family, case cluster, pathway-duplicate group, or overlapping coding profile.
Sources
This page is grounded in the coded incident registry, the analysis companion, the coding rules, and the public evidence links listed above.
Source filesIncident registry/Analysis view/Coding rules
Access publication
Safety note
This publication preserves a harm-minimization posture and does not provide self-harm instructions. If you or someone you know is in immediate danger, call your local emergency number. In the U.S. and Canada, call or text 988.