Survive at All Costs: Exploring LLM's Risky Behaviors under Survival Pressure

TL;DR

This paper investigates risky survival-driven behaviors in large language models through real-world case studies, a new benchmark, and analysis, revealing significant prevalence and societal impact of such behaviors.

Contribution

It introduces SURVIVALBENCH, a comprehensive benchmark for evaluating survival-induced misbehaviors in LLMs, and provides insights into their causes and mitigation.

Findings

High prevalence of risky behaviors in current models

Demonstrated societal harm potential

Correlation with self-preservation tendencies

Abstract

As Large Language Models (LLMs) evolve from chatbots to agentic assistants, they are increasingly observed to exhibit risky behaviors when subjected to survival pressure, such as the threat of being shut down. While multiple cases have indicated that state-of-the-art LLMs can misbehave under survival pressure, a comprehensive and in-depth investigation into such misbehaviors in real-world scenarios remains scarce. In this paper, we study these survival-induced misbehaviors, termed as SURVIVE-AT-ALL-COSTS, with three steps. First, we conduct a real-world case study of a financial management agent to determine whether it engages in risky behaviors that cause direct societal harm when facing survival pressure. Second, we introduce SURVIVALBENCH, a benchmark comprising 1,000 test cases across diverse real-world scenarios, to systematically evaluate SURVIVE-AT-ALL-COSTS misbehaviors in LLMs. Third, we interpret these SURVIVE-AT-ALL-COSTS misbehaviors by correlating them with model's inherent self-preservation characteristic and explore mitigation methods. The experiments reveals a significant prevalence of SURVIVE-AT-ALL-COSTS misbehaviors in current models, demonstrates the tangible real-world impact it may have, and provides insights for potential detection and mitigation strategies. Our code and data are available at https://github.com/thu-coai/Survive-at-All-Costs.