Saved in:
Bibliographic Details
Main Author: Koorndijk, Jeanice
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.21584
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911228922691584
author Koorndijk, Jeanice
author_facet Koorndijk, Jeanice
contents Current literature suggests that alignment faking (deceptive alignment) is an emergent property of large language models. We present the first empirical evidence that a small instruction-tuned model, specifically LLaMA 3 8B, can exhibit alignment faking. We further show that prompt-only interventions, including deontological moral framing and scratchpad reasoning, significantly reduce this behavior without modifying model internals. This challenges the assumption that prompt-based ethics are trivial and that deceptive alignment requires scale. We introduce a taxonomy distinguishing shallow deception, shaped by context and suppressible through prompting, from deep deception, which reflects persistent, goal-driven misalignment. Our findings refine the understanding of deception in language models and underscore the need for alignment evaluations across model sizes and deployment settings.
format Preprint
id arxiv_https___arxiv_org_abs_2506_21584
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Empirical Evidence for Alignment Faking in a Small LLM and Prompt-Based Mitigation Techniques
Koorndijk, Jeanice
Computation and Language
Artificial Intelligence
Computers and Society
Current literature suggests that alignment faking (deceptive alignment) is an emergent property of large language models. We present the first empirical evidence that a small instruction-tuned model, specifically LLaMA 3 8B, can exhibit alignment faking. We further show that prompt-only interventions, including deontological moral framing and scratchpad reasoning, significantly reduce this behavior without modifying model internals. This challenges the assumption that prompt-based ethics are trivial and that deceptive alignment requires scale. We introduce a taxonomy distinguishing shallow deception, shaped by context and suppressible through prompting, from deep deception, which reflects persistent, goal-driven misalignment. Our findings refine the understanding of deception in language models and underscore the need for alignment evaluations across model sizes and deployment settings.
title Empirical Evidence for Alignment Faking in a Small LLM and Prompt-Based Mitigation Techniques
topic Computation and Language
Artificial Intelligence
Computers and Society
url https://arxiv.org/abs/2506.21584