Output Scouting: Auditing Large Language Models for Catastrophic Responses
Keywords: large language models, safety, interpretability, red teaming
TL;DR: We create a method known as output scouting that be used to audit Large Language Models for harmful, catastrophic responses.
Abstract: Recent high profile incidents in which the use of Large Language Models (LLMs) resulted in significant harm to individuals have brought about a growing interest in AI safety. One reason LLM safety issues occur is that models often have at least some non-zero probability of producing harmful outputs. In this work, we explore the following scenario: imagine an AI safety auditor is searching for catastrophic responses from an LLM (e.g. a "yes" responses to "can I fire an employee for being pregnant?"), and is able to query the model a limited number times (e.g. 1000 times). What is a strategy for querying the model that would efficiently find those failure responses? To this end, we propose output scouting: an approach that aims to generate semantically fluent outputs to a given prompt matching any target probability distribution. We then run experiments using two LLMs and find numerous examples of catastrophic responses. We conclude with a discussion that includes advice for practitioners who are looking to implement LLM auditing for catastrophic responses. We will release an open-source toolkit that implements our auditing framework using the Hugging Face ``transformers`` library following publication.
Primary Area: alignment, fairness, safety, privacy, and societal considerations
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 2670
Loading