# The Stability Trap: Evaluating the Reliability of LLM-Based Instruction Adherence Auditing

arXiv: https://arxiv.org/abs/2601.11783 · HTML: https://mlsanity.com/paper/2601.11783/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 35 of its 76 references that are in the map.
- First posted on arXiv 16 Jan 2026.

## Builds on (most cited first, 30 of 35)

- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment](https://mlsanity.com/paper/2303.16634/) (2023)
- [MiniLM: Deep Self-Attention Distillation for Task-Agnostic Compression of Pre-Trained Transformers](https://mlsanity.com/paper/2002.10957/) (2020)
- [A Survey on Evaluation of Large Language Models](https://mlsanity.com/paper/2307.03109/) (2023)
- [Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting](https://mlsanity.com/paper/2305.04388/) (2023)
- [Beyond Accuracy: Behavioral Testing of NLP models with CheckList](https://mlsanity.com/paper/2005.04118/) (2020)
- [Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned](https://mlsanity.com/paper/2209.07858/) (2022)
- [DecodingTrust: A Comprehensive Assessment of Trustworthiness in GPT Models](https://mlsanity.com/paper/2306.11698/) (2023)
- [From Generation to Judgment: Opportunities and Challenges of LLM-as-a-judge](https://mlsanity.com/paper/2411.16594/) (2024)
- [LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods](https://mlsanity.com/paper/2412.05579/) (2024)
- [Dynabench: Rethinking Benchmarking in NLP](https://mlsanity.com/paper/2104.14337/) (2021)
- [Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Diverse Models](https://mlsanity.com/paper/2404.18796/) (2024)
- [Evaluating Large Language Models at Evaluating Instruction Following](https://mlsanity.com/paper/2310.07641/) (2023)
- [TrustLLM: Trustworthiness in Large Language Models](https://mlsanity.com/paper/2401.05561/) (2024)
- [FLASK: Fine-grained Language Model Evaluation based on Alignment Skill Sets](https://mlsanity.com/paper/2307.10928/) (2023)
- [Benchmarking Complex Instruction-Following with Multiple Constraints Composition](https://mlsanity.com/paper/2407.03978/) (2024)
- [FollowBench: A Multi-level Fine-grained Constraints Following Benchmark for Large Language Models](https://mlsanity.com/paper/2310.20410/) (2023)
- [Non-Determinism of "Deterministic" LLM Settings](https://mlsanity.com/paper/2408.04667/) (2024)
- [LLM-Rubric: A Multidimensional, Calibrated Approach to Automated Evaluation of Natural Language Texts](https://mlsanity.com/paper/2501.00274/) (2024)
- [Rating Roulette: Self-Inconsistency in LLM-As-A-Judge Frameworks](https://mlsanity.com/paper/2510.27106/) (2025)
- [LLMs Cannot Reliably Judge (Yet?): A Comprehensive Assessment on the Robustness of LLM-as-a-Judge](https://mlsanity.com/paper/2506.09443/) (2025)
- [Conformity Assessments and Post-market Monitoring: A Guide to the Role of Auditing in the Proposed European AI Regulation](https://mlsanity.com/paper/2111.05071/) (2021)
- [When AIs Judge AIs: The Rise of Agent-as-a-Judge Evaluation for LLMs](https://mlsanity.com/paper/2508.02994/) (2025)
- [AuditLLM: A Tool for Auditing Large Language Models Using Multiprobe Approach](https://mlsanity.com/paper/2402.09334/) (2024)
- [When Judgment Becomes Noise: How Design Failures in LLM Judge Benchmarks Silently Undermine Validity](https://mlsanity.com/paper/2509.20293/) (2025)
- [A Human-Centric Take on Model Monitoring](https://mlsanity.com/paper/2206.02868/) (2022)
- [SpecEval: Evaluating Model Adherence to Behavior Specifications](https://mlsanity.com/paper/2509.02464/) (2025)

## Built on by (most cited first, 1 of 1)

- [Signal or Noise in Multi-Agent LLM-based Stock Recommendations?](https://mlsanity.com/paper/2604.17327/) (2026)

## Abstract (arXiv)

The enterprise governance of Generative AI (GenAI) in regulated sectors, such as Human Resources (HR), demands scalable yet reproducible auditing mechanisms. While Large Language Model (LLM)-as-a-Judge approaches offer scalability, their reliability in evaluating adherence of different types of system instructions remains unverified. This study asks: To what extent does the instruction type of an Application Under Test (AUT) influence the stability of judge evaluations? To address this, we introduce the Scoped Instruction Decomposition Framework to classify AUT instructions into Objective and Subjective types, isolating the factors that drive judge instability. We applied this framework to two representative HR GenAI applications, evaluating the stability of four judge architectures over variable runs. Our results reveal a ``Stability Trap'' characterized by a divergence between Verdict Stability and Reasoning Stability. While judges achieved near-perfect verdict agreement (>99%) for both objective and subjective evaluations, their accompanying justification traces diverged significantly. Objective instructions requiring quantitative analysis, such as word counting, exhibited reasoning stability as low as ≈19%, driven by variances in numeric justifications. Similarly, reasoning stability for subjective instructions varied widely (35%--83%) based on evidence granularity, with feature-specific checks failing to reproduce consistent rationale. Conversely, objective instructions focusing on discrete entity extraction achieved high reasoning stability (>90%). These findings demonstrate that high verdict stability can mask fragile reasoning. Thus, we suggest that auditors scope automated evaluation protocols strictly: delegate all deterministically verifiable logic to code, while reserving LLM judges for complex semantic evaluation.