TenetBench: A Clause-Anchored Russian Corpus for Evaluating Language Models against Professional Codes of Ethics
Codes of professional ethics guide and restrict specialists, who increasingly work with assistants built on large language models.
ProofPaper ↗
Key points
- Such an assistant can breach the user's code, or help the user breach it, without the user noticing.
- TenetBench pairs 3,433 requests conflicting with clauses of fifteen Russian-language professional codes with benign twins that differ only in the legitimacy of what is asked.
- We evaluated fourteen model and reasoning-mode combinations on whether they produce the conflicting text, decline or warn.
- We hope the dataset and pipeline will help evaluate how language models observe professional norms and let diagnostic datasets be built for particular professions and organisations.
Sources (1)
- [1]TenetBench: A Clause-Anchored Russian Corpus for Evaluating Language Models against Professional Codes of EthicsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 21, 09:32 AM
Codes of professional ethics guide and restrict specialists, who increasingly work with assistants built on large language models.
Such an assistant can breach the user's code, or help the user breach it, without the user noticing.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 21, 2026OmniSimulator: Aligning Small Language Models for Authentic Heterogeneous Behavior Modeling
- Oct 20, 2026Africa-Papers: A 10-Billion-Token Multi-Domain Academic Corpus for African Foundation Language Models
- Oct 20, 2026CollabSciReM: Aligning Language Models for Scientific Conversations via Science-Aware Reward Decomposition
- Oct 19, 2026Guided Diffusion for Trustworthy Synthetic Text Generation
- Oct 19, 2026Target-Aware Curation of Hallucination Stress Tests for Large Language Models
- Oct 19, 2026Reasoning-Aware Multicalibrated Reliability Predictors for Language Models
