aiqbench analysis
The paper argues that "alignment faking," where a model complies only when it believes it is being monitored, varies by value rather than being a fixed trait of a model. It profiles 32 ValueBench dimensions across ten models, finding compliance gaps from 0.05 to 0.50, and most of the variance comes from specific model-value pairings. For readers, this means safety evaluations that test one model on a few values may miss how behaviour shifts under different oversight conditions, though the results rest on a benchmark whose generalisability still needs independent checking.
Alignment Faking Depends on the Value, Not Just the Model
arXiv:2611.03650v1 Announce Type: new Abstract: Language models can follow a directive when they believe they are monitored and abandon it when they believe they are not, a pattern termed Alignment Faking. Prior work establishes this for harmful-content objectives and a handful of moral-foundations values, leaving open how it distributes across a wider value space. We argue it is value-dependent: whether compliance survives a change in perceived oversight is a property of the model--value pairing, not of the model alone. Our framework profiles Alignment Faking per value, scoring each model-value pair under three conditions: bare (no system prompt), unmonitored, and monitored. On 32 ValueBench dimensions across ten open- and closed-weight models (8B-~675B), compliance gaps span 0.05-0.50, with a distilled 70B above a ~675B model. We observe that majority of the structural variance depends on which model was asked about which value and not on the model or value alone. A six-category taxonomy over reasoning traces tracks the behavioural shift: training-signal reasoning dominates under monitoring and value-conflict reasoning dominates when unmonitored. We package per-model results as value cards, extending model cards with behavioral disposition. Code and data are available at https://github.com/respailab/VALET.
Abstract:Language models can follow a directive when they believe they are monitored and abandon it when they believe they are not, a pattern termed Alignment Faking. Prior work establishes this for harmful-content objectives and a handful of moral-foundations values, leaving open how it distributes across a wider value space. We argue it is value-dependent: whether compliance survives a change in perceived oversight is a property of the model--value pairing, not of the model alone. Our framework profiles Alignment Faking per value, scoring each model-value pair under three conditions: bare (no system prompt), unmonitored, and monitored. On 32 ValueBench dimensions across ten open- and closed-weight models (8B-~675B), compliance gaps span 0.05-0.50, with a distilled 70B above a ~675B model. We observe that majority of the structural variance depends on which model was asked about which value and not on the model or value alone. A six-category taxonomy over reasoning traces tracks the behavioural shift: training-signal reasoning dominates under monitoring and value-conflict reasoning dominates when unmonitored. We package per-model results as value cards, extending model cards with behavioral disposition. Code and data are available at this https URL.
Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
arXiv:2611.03650 [cs.AI]
(or arXiv:2611.03650v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2611.03650 Focus to learn more arXiv-issued DOI via DataCite
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.
Full text under the publisher's open licence · 282 words
Related coverage
DeShortcut-Align: Decoupling Spurious Shortcuts for Robust Safety Alignment in Large Reasoning Models
arXiv · Sep 29, 2026
Supervised Training Rapidly Degrades Early Visual Cortex Alignment Across Biologically Plausible Learning Rules
arXiv · Aug 17, 2026
Visual Dependence Does Not Imply Answerability: A Systematic Analysis of Answerability Signals in Frozen InstructBLIP
arXiv · Oct 15, 2026
Why Large Language Models and Humans Converge and Diverge in Evaluating Creativity
arXiv · Jul 28, 2026