Saved in:
Bibliographic Details
Main Authors: Blandfort, Phil, Graham, Robert
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2511.00554
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914130067193856
author Blandfort, Phil
Graham, Robert
author_facet Blandfort, Phil
Graham, Robert
contents Activation probes are attractive monitors for AI systems due to low cost and latency, but their real-world robustness remains underexplored. We ask: What failure modes arise under realistic, black-box adversarial pressure, and how can we surface them with minimal effort? We present a lightweight black-box red-teaming procedure that wraps an off-the-shelf LLM with iterative feedback and in-context learning (ICL), and requires no fine-tuning, gradients, or architectural access. Running a case study with probes for high-stakes interactions, we show that our approach can help discover valuable insights about a SOTA probe. Our analysis uncovers interpretable brittleness patterns (e.g., legalese-induced FPs; bland procedural tone FNs) and reduced but persistent vulnerabilities under scenario-constraint attacks. These results suggest that simple prompted red-teaming scaffolding can anticipate failure patterns before deployment and might yield promising, actionable insights to harden future probes.
format Preprint
id arxiv_https___arxiv_org_abs_2511_00554
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Red-teaming Activation Probes using Prompted LLMs
Blandfort, Phil
Graham, Robert
Machine Learning
Artificial Intelligence
Activation probes are attractive monitors for AI systems due to low cost and latency, but their real-world robustness remains underexplored. We ask: What failure modes arise under realistic, black-box adversarial pressure, and how can we surface them with minimal effort? We present a lightweight black-box red-teaming procedure that wraps an off-the-shelf LLM with iterative feedback and in-context learning (ICL), and requires no fine-tuning, gradients, or architectural access. Running a case study with probes for high-stakes interactions, we show that our approach can help discover valuable insights about a SOTA probe. Our analysis uncovers interpretable brittleness patterns (e.g., legalese-induced FPs; bland procedural tone FNs) and reduced but persistent vulnerabilities under scenario-constraint attacks. These results suggest that simple prompted red-teaming scaffolding can anticipate failure patterns before deployment and might yield promising, actionable insights to harden future probes.
title Red-teaming Activation Probes using Prompted LLMs
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2511.00554