aiqbench analysis
This research reveals that nearly all evaluated LLMs exhibit significant context sensitivity in moral judgments, often shifting toward rule-violating behaviors when specific contextual cues are introduced. The study highlights a critical misalignment between models and humans regarding which contextual variations trigger these shifts, suggesting that base-case alignment does not guarantee consistent ethical behavior across varying scenarios. By demonstrating an activation steering technique to control this sensitivity, the work offers a concrete method for improving the robustness of AI moral reasoning in dynamic environments.
On the Context Sensitivity of LLM Moral Judgment
arXiv:2603.23114v2 Announce Type: replace-cross Abstract: A human's moral decision depends heavily on the context. Yet research on LLM morality has largely studied fixed scenarios. We address this gap by introducing Contextual MoralChoice, a dataset of moral dilemmas with systematic contextual variations known from moral psychology to shift human judgment: consequentialist, emotional, and relational. Evaluating 22 LLMs, we find that nearly all models are context-sensitive, shifting their judgments towards rule-violating behavior. Comparing with a human survey, we find that models and humans are most triggered by different contextual variations, and that a model aligned with human judgments in the base case is not necessarily aligned in its contextual sensitivity. This raises the question of controlling contextual sensitivity, which we address with an activation steering approach that can reliably increase or decrease a model's contextual sensitivity. Code and data: https://github.com/adrian-sauter/contextual_moralchoice.
Abstract:A human's moral decision depends heavily on the context. Yet research on LLM morality has largely studied fixed scenarios. We address this gap by introducing Contextual MoralChoice, a dataset of moral dilemmas with systematic contextual variations known from moral psychology to shift human judgment: consequentialist, emotional, and relational. Evaluating 22 LLMs, we find that nearly all models are context-sensitive, shifting their judgments towards rule-violating behavior. Comparing with a human survey, we find that models and humans are most triggered by different contextual variations, and that a model aligned with human judgments in the base case is not necessarily aligned in its contextual sensitivity. This raises the question of controlling contextual sensitivity, which we address with an activation steering approach that can reliably increase or decrease a model's contextual sensitivity. Code and data: this https URL.
Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
arXiv:2603.23114 [cs.AI]
(or arXiv:2603.23114v2 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2603.23114 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 · 238 words
Related coverage
The iVAIS Manifesto: Safety Through Character, Not Compliance
LessWrong · Jul 30, 2026
The basic argument for making AI better at moral philosophy
LessWrong · Oct 14, 2026
The Continual Constitutional Convention: How language models revise the documents that govern them
AI Alignment Forum · Oct 13, 2026
Morality is Contextual: Learning Interpretable Moral Contexts from Human Data with Probabilistic Clustering and Large Language Models
arXiv · Oct 6, 2026