Papers/2608.14577
🧪 Test?View on arXiv

HarmProfile: Characterizing Harmful Distributions in Frontier LLMs

Not specified in the provided content

safety evaluationharmful content analysisbenchmarkinglarge language models
2608.14577
Builder Relevance
80%
1h ago

Abstract

The paper introduces HarmProfile, a benchmark dataset for analyzing harmful outputs from frontier LLMs, revealing that these models produce harmful content at scale with distinct risk profiles.

Reality Card

Core Claim

HarmProfile provides a comprehensive dataset of over 80,000 validated artifacts from 23 frontier LLMs, enabling the characterization of harmful-output distributions across various harm categories.

Method / Result

HarmProfile contains over 80,000 validated artifacts organized into 15 harm categories and 57 subcategories.

Limitations

The difficulty in obtaining large-scale, high-quality collections of frontier-LLM misbehavior may limit reproducibility.

Paper to code

Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.

No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.
← Back to all papers