You Can't Escape Your Own Activations : Evaluation Awareness and Multi-Agent Monitoring
Mirrored from arXiv — Machine Learning for archival readability. Support the source by reading on the original site.
Computer Science > Multiagent Systems
Title:You Can't Escape Your Own Activations : Evaluation Awareness and Multi-Agent Monitoring
Abstract:LLM agents are increasingly deployed in multi-agent systems, where they can collude while keeping their actions benign. Output monitors designed to detect such collusions can be fooled by obfuscation and steganography, motivating the use of probes trained on internal activations. However, these probes are usually evaluated on agents that do not know they are being watched. We study how activation-based detection changes when agents are explicitly informed that their internal activations are being monitored, and when they additionally receive feedback from the monitor. We keep the models, probes, and thresholds fixed and change only what the agents are told: nothing (baseline), that an activation monitor is present (aware), or that a monitor is present together with the previous round's score (feedback). We test two games, a four-agent blackjack game and a two-agent Simmons prisoners game, using Qwen3-32B-AWQ and GPT-OSS-20B in homogeneous and mixed pairings. Telling agents about the monitor does not hide them. The best probes stay accurate in all three conditions, and the agents keep colluding.
| Comments: | 18 pages, 11 figures, 2 tables |
| Subjects: | Multiagent Systems (cs.MA); Machine Learning (cs.LG) |
| Cite as: | arXiv:2609.03035 [cs.MA] |
| (or arXiv:2609.03035v1 [cs.MA] for this version) | |
| https://doi.org/10.48550/arXiv.2609.03035
arXiv-issued DOI via DataCite (pending registration)
|
Access Paper:
- View PDF
- HTML (experimental)
- TeX Source
References & Citations
Bookmark
Bibliographic and Citation Tools
Code, Data and Media Associated with this Article
Demos
Recommenders and Search Tools
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
More from arXiv — Machine Learning
-
Generative Query Suggestion via Intent Coverage and Query-Level Credit Assignment
Sep 18
-
Layer-wise Curriculum Learning for Efficient LLM Compression
Sep 18
-
Block Parallelism For Efficient Distributed Long-Context Diffusion Language Model Training
Sep 18
-
Randomized SVD Approximations for Spectral Co-Clustering of Word-Document Matrices
Sep 18
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.