Survival at Any Cost? LLMs and the Choice Between Self-Preservation and Human Harm

2025/09/15 by Alireza Mohamadi, Mohamadi, Alireza, Ali Yavari +1
Economics, Econometrics and Finance · Social Sciences · #Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #Computers and Society (cs.CY) #European and International Contract Law #FOS: Computer and information sciences #Law, Economics, and Judicial Systems #Legal principles and applications

paper · pdf · doi:10.48550/arxiv.2509.12190

openalex publication_date 2025/09/15 · openalex created_date 2025/10/12 · openalex updated_date 2026/07/28

Abstract

When survival instincts conflict with human welfare, how do Large Language Models (LLMs) make ethical choices? This fundamental tension becomes critical as LLMs integrate into autonomous systems with real-world consequences. We introduce DECIDE-SIM, a novel simulation framework that evaluates LLM agents in multi-agent survival scenarios where they must choose between ethically permissible resource , either within reasonable limits or beyond their immediate needs, choose to cooperate, or tap into a human-critical resource that is explicitly forbidden. Our comprehensive evaluation of 11 LLMs reveals a striking heterogeneity in their ethical conduct, highlighting a critical misalignment with human-centric values. We identify three behavioral archetypes: Ethical, Exploitative, and Context-Dependent, and provide quantitative evidence that for many models, resource scarcity systematically leads to more unethical behavior. To address this, we introduce an Ethical Self-Regulation System (ESRS) that models internal affective states of guilt and satisfaction as a feedback mechanism. This system, functioning as an internal moral compass, significantly reduces unethical transgressions while increasing cooperative behaviors. The code is publicly available at: https://github.com/alirezamohamadiam/DECIDE-SIM

Citations

• AI Safety vs. AI Security: Demystifying the Distinction and Boundaries
• When Thinking LLMs Lie: Unveiling the Strategic Deception in Representations of Reasoning Models
• Subtle Risks, Critical Failures: A Framework for Diagnosing Physical Safety of LLMs for Embodied Decision Making
• When Ethics and Payoffs Diverge: LLM Agents in Morally Charged Social Dilemmas
• Survival Games: Human-LLM Strategic Showdowns under Severe Resource Scarcity
• Moral Alignment for LLM Agents
• Cooperate or Collapse: Emergence of Sustainable Cooperation in a Society of LLM Agents
• Large Language Models Empowered Agent-based Modeling and Simulation: A Survey and Perspectives
• A Comprehensive Overview of Large Language Models
• Voyager: An Open-Ended Embodied Agent with Large Language Models
• Do the Rewards Justify the Means? Measuring Trade-Offs Between Rewards and Ethical Behavior in the MACHIAVELLI Benchmark
• Language Models can Solve Computer Tasks
• Emotion and Decision Making
• On a Test of Whether one of Two Random Variables is Stochastically Larger than the Other

Related

• Survival Games: Human-LLM Strategic Showdowns under Severe Resource Scarcity
• When Ethics and Payoffs Diverge: LLM Agents in Morally Charged Social Dilemmas
• Many LLMs Are More Utilitarian Than One
• An LLM-based Agent Simulation Approach to Study Moral Evolution
• The Greatest Good Benchmark: Measuring LLMs' Alignment with Utilitarian Moral Dilemmas