{
  "abstract": "Chain-of-thought (CoT) monitoring aims to detect potentially harmful behaviors in large language models (LLMs) from their CoTs. Intervention-based evaluations measure this property by asking whether a monitor can detect the role of an intervention from the CoT when the targeted behavior occurs, typically using minimal-criterion sensitivity ($\\mathsf{TPR_{mc}}$) proposed by~\\citep{guan2025monitoring}. However, a target behavior observed under intervention is not necessarily caused by it, and we show that this metric can overestimate how often a monitor detects target behaviors actually caused by the intervention. Specifically, we establish $\\mathsf{TPR_{mc}}$ as an upper bound on this causal detection rate and derive a complementary lower bound, which provides a tighter evaluation framework for CoT monitorability. With this framework, we study whether LLMs can be trained to make their CoTs easier to monitor. We first construct a training dataset of paired control--intervention tasks spanning multiple intervention types. Using the dataset, we design and apply three intervention-specific training methods based on off-policy self-distillation, reinforcement learning, and on-policy self-distillation to train separate variants of each base model. Across various model-monitor configurations, all three methods generally improve CoT monitorability over the corresponding base models under our extended evaluation framework. Our code is available at https://github.com/xinyue-lxy/learning-to-be-monitored.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2610.31733",
  "author_total": 4,
  "authors": [
    {
      "name": "Qiang Fu"
    },
    {
      "name": "Xinyue Li"
    },
    {
      "name": "Andre Wibisono"
    },
    {
      "name": "Arman Cohan"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/2r729ck798qpg.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2610.31733",
  "fulltext_url": null,
  "id": "2r729ck798qpg",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2610.31733",
  "language": null,
  "licence": "LicenseRef-arXiv-nonexclusive-1.0",
  "licence_label": "arXiv licence",
  "licence_source": "article",
  "licence_url": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.AI"
    ],
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "stat.ML"
    ],
    [
      "openalex",
      "1702|Explainable Artificial Intelligence (XAI)"
    ]
  ],
  "openalex": "W7223692198",
  "pdf_url": "https://arxiv.org/pdf/2610.31733",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-22",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/2r729ck798qpg.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:35:25Z",
      "native_id": "2610.31733",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2610.31733"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.ML",
      "subfield": "statistics"
    }
  ],
  "title": "Learning to Be Monitored: Causal Bounds and Intervention-Aware Post-Training Recipes for Chain-of-Thought Monitorability",
  "title_text": "Learning to Be Monitored: Causal Bounds and Intervention-Aware Post-Training Recipes for Chain-of-Thought Monitorability",
  "type": "preprint",
  "updated_at": "2026-10-30T12:35:25Z",
  "url": "https://research.draconitic.com/paper/2r729ck798qpg",
  "venue": "arXiv",
  "versions": []
}