Evaluating Proactive Risk Awareness of Large Language Models

TL;DR

This paper introduces a framework and dataset to evaluate whether large language models can proactively identify and warn about potential ecological harms before they occur, highlighting gaps in current safety measures.

Contribution

It presents a novel proactive risk awareness evaluation framework and the Butterfly dataset for ecological risk assessment in LLMs, revealing significant safety gaps.

Findings

LLMs show reduced proactive awareness with response length restrictions

Cross-lingual similarities in risk awareness were observed

Persistent blind spots in species protection across models

Abstract

As large language models (LLMs) are increasingly embedded in everyday decision-making, their safety responsibilities extend beyond reacting to explicit harmful intent toward anticipating unintended but consequential risks. In this work, we introduce a proactive risk awareness evaluation framework that measures whether LLMs can anticipate potential harms and provide warnings before damage occurs. We construct the Butterfly dataset to instantiate this framework in the environmental and ecological domain. It contains 1,094 queries that simulate ordinary solution-seeking activities whose responses may induce latent ecological impact. Through experiments across five widely used LLMs, we analyze the effects of response length, languages, and modality. Experimental results reveal consistent, significant declines in proactive awareness under length-restricted responses, cross-lingual similarities, and persistent blind spots in (multimodal) species protection. These findings highlight a critical gap between current safety alignment and the requirements of real-world ecological responsibility, underscoring the need for proactive safeguards in LLM deployment.