# When Compliance Data Masquerades as Evaluation: Measurement Validity for Deployed AI Systems

arXiv: https://arxiv.org/abs/2609.13642 · HTML: https://mlsanity.com/paper/2609.13642/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 24 of its 45 references that are in the map.
- First posted on arXiv 12 Sep 2026.

## Builds on (most cited first, 24 of 24)

- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Concrete Problems in AI Safety](https://mlsanity.com/paper/1606.06565/) (2016)
- [Do ImageNet Classifiers Generalize to ImageNet?](https://mlsanity.com/paper/1902.10811/) (2019)
- [Datasheets for Datasets](https://mlsanity.com/paper/1803.09010/) (2018)
- [WILDS: A Benchmark of in-the-Wild Distribution Shifts](https://mlsanity.com/paper/2012.07421/) (2020)
- [Model Cards for Model Reporting](https://mlsanity.com/paper/1810.03993/) (2018)
- [Dynabench: Rethinking Benchmarking in NLP](https://mlsanity.com/paper/2104.14337/) (2021)
- [Closing the AI Accountability Gap: Defining an End-to-End Framework for Internal Algorithmic Auditing](https://mlsanity.com/paper/2001.00973/) (2020)
- [AI and the Everything in the Whole Wide World Benchmark](https://mlsanity.com/paper/2111.15366/) (2021)
- [Measurement and Fairness](https://mlsanity.com/paper/1912.05511/) (2019)
- [Data and its (dis)contents: A survey of dataset development and use in machine learning research](https://mlsanity.com/paper/2012.05345/) (2020)
- [What Will it Take to Fix Benchmarking in Natural Language Understanding?](https://mlsanity.com/paper/2104.02145/) (2021)
- [Measuring what Matters: Construct Validity in Large Language Model Benchmarks](https://mlsanity.com/paper/2511.04703/) (2025)
- [Position: Evaluating Generative AI Systems Is a Social Science Measurement Challenge](https://mlsanity.com/paper/2502.00561/) (2025)
- [Waymo Public Road Safety Performance Data](https://mlsanity.com/paper/2011.00038/) (2020)
- [CEBench: A Benchmarking Toolkit for the Cost-Effectiveness of LLM Pipelines](https://mlsanity.com/paper/2407.12797/) (2024)
- [Comparison of Waymo Rider-Only Crash Data to Human Benchmarks at 7.1 Million Miles](https://mlsanity.com/paper/2312.12675/) (2023)
- [SouLLMate: An Application Enhancing Diverse Mental Health Support with Adaptive LLMs, Prompt Engineering, and RAG Techniques](https://mlsanity.com/paper/2410.16322/) (2024)
- [Advancing Mental Health Pre-Screening: A New Custom GPT for Psychological Distress Assessment](https://mlsanity.com/paper/2408.01614/) (2024)
- [RAVE Checklist: Recommendations for Overcoming Challenges in Retrospective Safety Studies of Automated Driving Systems](https://mlsanity.com/paper/2408.07758/) (2024)
- [Attacking LLMs and AI Agents: Advertisement Embedding Attacks Against Large Language Models](https://mlsanity.com/paper/2508.17674/) (2025)
- [AquaSentinel: Next-Generation AI System Integrating Sensor Networks for Urban Underground Water Pipeline Anomaly Detection via Collaborative MoE-LLM Agent Architecture](https://mlsanity.com/paper/2511.15870/) (2025)
- [RPKT: Learning What You Don't -- Know Recursive Prerequisite Knowledge Tracing in Conversational AI Tutors for Personalized Learning](https://mlsanity.com/paper/2508.11892/) (2025)
- [Spatio-Temporal Graph Unlearning](https://mlsanity.com/paper/2511.09404/) (2025)

## Abstract (arXiv)

We argue that a recurring failure in the evaluation of deployed AI systems occurs when data collected for operational monitoring or regulatory compliance are interpreted as if they were designed for comparative evaluation. Automated driving provides a concrete example of this problem. U.S. disengagement and crash-reporting regimes produce valuable operational evidence, but differences in reporting scope, exposure, deployment domain, event capture, and comparator construction limit the safety claims that can be supported from these measurements alone. We frame this issue as a measurement-validity problem in AI evaluation rather than as a transportation-specific data limitation. We argue that comparative claims about deployed AI systems require alignment between the intended capability, measured outcome, exposure opportunity, deployment domain, data-generation process, and evaluation comparator. Using automated-driving safety evaluation as a case study, we propose an evaluation contract that makes these assumptions explicit before operational data are interpreted as evidence of comparative performance. The broader implication is that data useful for monitoring deployed AI systems are not automatically valid benchmarks for evaluating them.