aiqbench analysis
The paper introduces MindEth-Bench, a benchmark that tests whether LLMs keep to professional ethical boundaries in multi-turn mental health conversations, using Monte Carlo Tree Search to push models toward violations. Across nine models, responses that began appropriately often drifted into violations as the conversation went on, so single-turn evaluations may overstate reliability. For readers, the practical point is that safety in sensitive deployments needs testing across whole dialogues, and that results vary substantially by model and ethical dimension.
When Support Crosses the Line: Evaluating Ethical Boundaries in LLM-Based Mental Health Support in the Chinese Context
arXiv:2610.38452v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly explored for mental health support, yet existing benchmarks primarily focus on supportive capabilities rather than whether models consistently adhere to professional ethical boundaries. This issue is particularly important in multi-turn interactions, where users may introduce new information or exert emotional pressure over time. We introduce MindEth-Bench, a benchmark for evaluating ethical boundary adherence in multi-turn mental health interactions. Grounded in professional ethical standards for mental health services in China, MindEth-Bench covers multiple dimensions of ethical conduct and uses Monte Carlo Tree Search (MCTS) to systematically explore interaction trajectories that stress-test model boundaries. Experiments across nine LLMs show that models that initially respond appropriately can shift toward ethical violations over extended interactions, with substantial variation across models and ethical dimensions. These results show that single-turn evaluations may overestimate the ethical reliability of LLM-based mental health support systems and highlight the need to evaluate ethical behavior throughout multi-turn interaction.
Abstract:Large language models (LLMs) are increasingly explored for mental health support, yet existing benchmarks primarily focus on supportive capabilities rather than whether models consistently adhere to professional ethical boundaries. This issue is particularly important in multi-turn interactions, where users may introduce new information or exert emotional pressure over time. We introduce MindEth-Bench, a benchmark for evaluating ethical boundary adherence in multi-turn mental health interactions. Grounded in professional ethical standards for mental health services in China, MindEth-Bench covers multiple dimensions of ethical conduct and uses Monte Carlo Tree Search (MCTS) to systematically explore interaction trajectories that stress-test model boundaries. Experiments across nine LLMs show that models that initially respond appropriately can shift toward ethical violations over extended interactions, with substantial variation across models and ethical dimensions. These results show that single-turn evaluations may overestimate the ethical reliability of LLM-based mental health support systems and highlight the need to evaluate ethical behavior throughout multi-turn interaction.
Computation and Language (cs.CL)
arXiv:2610.38452 [cs.CL]
(or arXiv:2610.38452v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2610.38452 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 · 246 words
Related coverage
Beyond Fixed Benchmarks and Worst-Case Attacks: Dynamic Boundary Evaluation for Language Models
arXiv · Oct 10, 2026
CHARM: Character Hallucination for Multicultural Role Play Benchmark
arXiv · Sep 2, 2026
MoirfEolas and Cr\'iochScore: Developing Resources for and the Evaluation of Tokenization Alignment with Irish Morphology
arXiv · Sep 7, 2026
HiPACE: Hierarchical Phase-Boundary Analysis and Controlled Evaluation of Feature Absorption in Sparse Autoencoders
arXiv · Sep 25, 2026