Metraipolis › archive › dc_cd6ad404264c

Multi-Agent Collusion and the J-Space — arXiv:2611.02153

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.02153
License (as archived)
CC0-1.0
Archived
2026-11-03T18:23:34.550Z
SHA-256 of the archived text
3875db96969b1b981823933458dadf0004f306738f3e601eb053a58764524322
Stable URL of this page
https://metraipolis.com/documento/dc_cd6ad404264c

How to cite

Multi-Agent Collusion and the J-Space — arXiv:2611.02153 — arXiv (en), https://arxiv.org/abs/2611.02153 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cd6ad404264c (Metraipolis archive dc_cd6ad404264c, sha256 3875db96969b1b981823933458dadf0004f306738f3e601eb053a58764524322)
BibTeX
@misc{dccd6ad404264c,
  title = {Multi-Agent Collusion and the J-Space — arXiv:2611.02153},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.02153},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cd6ad404264c; sha256 3875db96969b1b981823933458dadf0004f306738f3e601eb053a58764524322},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.02153. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

LLM agents are increasingly deployed in Multi-Agent Systems (MAS), and they introduce the risk of coordination among agents that might escape human oversight. The J-space has been used to uncover deception in single LLMs, but its use for uncovering and steering collusion between agents remains unexplored. We show, with interpretability experiments on DeepSeek-V4-Flash (284B) on the NARCBench collusion benchmark, that colluding agents' J-space holds the collusion itself: before replying, 86% of colluders represent misrepresentation, concealment or remorse. In a steganographic card-counting task, the receiver's J-space holds a hidden count, and causal interventions alter its count. We also introduce the Jacobian Pair Contrast (J-PACT), an unsupervised method for detecting collusion between agents. We evaluate J-PACT on four open-weight models: GPT-OSS-20B, Llama-3.3-70B-IT, Qwen3-32B, and DeepSeek-V4-Flash (284B). J-PACT retains 85% of the above-chance AUROC of supervised NARCBench probes, matches them on steganographic scenarios, and names the colluding pair up to its complement in 99% of detected runs.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2611.02153 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-11-03T18:10:07.956Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).