Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages

[2607.06596] Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages Skip to main content Search Submit Donate Log in Computer Science > Cryptography and Security arXiv:2607.06596 (cs) [Submitted on 6 Jul 2026 ( v1 ), last revised 25 Aug 2026 (this version, v3)] Title: Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages Authors: Lucas Pinto View PDF HTML (experimental) Abstract: Trusted monitoring is a central defense in AI control: a cheaper trusted model scores an untrusted model's actions for sabotage, and the most suspicious are audited or deferred. Such monitors are evaluated against one or two untrusted models, and the accuracy is reported as a property of the monitor. We ask whether it is partly a property of the pairing. We make the untrusted policy family the controlled axis: we fit a monitor on family A's transcripts, apply it to family B, and decompose the cross-family AUROC into how obvious each family's sabotage is, how capable each monitor is, and the residual own-family advantage after both are removed: the interaction. On code-backdoor transcripts the interaction is positive and survives the dominant confounds: +0.172 (95% CI [+0.158, +0.185]) on four open-weight families on a strict leak-free basis, holding across an eight-family matrix spanning four frontier vendors, blatancy-matched controls, a trained transcript-reading monitor up to 34B parameters, and a second code benchmark. The gap is geometric: each lineage's sabotage occupies a different scoring direction, and rotating the monitor's scoring direction toward the target lineage recovers most of the lost detection while a matched random rotation recovers nothing. At a 1% audit budget under defer-to-trusted, an off-lineage monitor catches 19% of attack attempts where an in-lineage monitor catches 41% at the same audit cost, a +21.7-point gap. Control evaluations should report cross-family transfer matrices: a single-pairing accuracy overstates the safety a monitor delivers against a lineage it was not calibrated on. We give a four-step protocol that acts on the gap, with each step a measured result. Comments: 28 pages, 11 figures. Accepted at the COLM 2026 AI Measurement Science workshop. Code: this https URL Subjects: Cryptography and Security (cs.CR) ; Machine Learning (cs.LG) Cite as: arXiv:2607.06596 [cs.CR] (or arXiv:2607.06596v3 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2607.06596 Focus to learn more arXiv-issued DOI via DataCite Submission history From: Lucas Pinto [ view email ] [v1] Mon, 6 Jul 2026 22:40:07 UTC (1,372 KB) [v2] Mon, 20 Jul 2026 21:34:34 UTC (1,373 KB) [v3] Tue, 25 Aug 2026 06:16:31 UTC (1,373 KB) Full-text links: Access Paper: View PDF HTML (experimental) TeX Source view license Current browse context: cs.CR < prev | next > new | recent | 2026-07 Change to browse by: cs cs.LG References & Citations NASA ADS Google Scholar Semantic Scholar export BibTeX citation Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer ( What is the Explorer? ) Connected Papers Toggle Connected Papers ( What is Connected Papers? ) Litmaps Toggle Litmaps ( What is Litmaps? ) scite.ai Toggle scite Smart Citations ( What are Smart Citations? ) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv ( What is alphaXiv? ) Links to Code Toggle CatalyzeX Code Finder for Papers ( What is CatalyzeX? ) DagsHub Toggle DagsHub ( What is DagsHub? ) GotitPub Toggle Gotit.pub ( What is GotitPub? ) Huggingface Toggle Hugging Face ( What is Huggingface? ) ScienceCast Toggle ScienceCast ( What is ScienceCast? ) Demos Demos Replicate Toggle Replicate ( What is Replicate? ) Spaces Toggle Hugging Face Spaces ( What is Spaces? ) Spaces Toggle TXYZ.AI ( What is TXYZ.AI? ) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower ( What are Influence Flowers? ) Core recommender toggle CORE Recommender ( What is CORE? ) Author Venue Institution Topic About arXivLabs 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 . Which authors of this paper are endorsers? | Disable MathJax ( What is MathJax? ) We gratefully acknowledge support from our major funders , member institutions , and all contributors. About · Help · Contact · Subscribe · Copyright · Privacy · Accessibility · Operational Status (opens in new tab) Major funding support from

Related documents

• Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages#670671
• Gram: Assessing sabotage propensities via automated alignment auditing#238599
• Vigil: Accountable Liveness against Selective Silence#1012942
• Vigil: Accountable Liveness against Selective Silence#965379
• Inheritance entropy: A model-independent method to probe the hereditary structure of cell lineage trees#1001864