Saved in:
Bibliographic Details
Main Authors: Yu, Cheng, Engelmann, Severin, Cao, Ruoxuan, Ali, Dalia, Papakyriakopoulos, Orestis
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.23112
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918326555377664
author Yu, Cheng
Engelmann, Severin
Cao, Ruoxuan
Ali, Dalia
Papakyriakopoulos, Orestis
author_facet Yu, Cheng
Engelmann, Severin
Cao, Ruoxuan
Ali, Dalia
Papakyriakopoulos, Orestis
contents AI safety benchmarks are pivotal for safety in advanced AI systems; however, they have significant technical, epistemic, and sociotechnical shortcomings. We present a review of 210 safety benchmarks that maps out common challenges in safety benchmarking, documenting failures and limitations by drawing from engineering sciences and long-established theories of risk and safety. We argue that adhering to established risk management principles, mapping the space of what can(not) be measured, developing robust probabilistic metrics, and efficiently deploying measurement theory to connect benchmarking objectives with the world can significantly improve the validity and usefulness of AI safety benchmarks. The review provides a roadmap on how to improve AI safety benchmarking, and we illustrate the effectiveness of these recommendations through quantitative and qualitative evaluation. We also introduce a checklist that can help researchers and practitioners develop robust and epistemologically sound safety benchmarks. This study advances the science of benchmarking and helps practitioners deploy AI systems more responsibly.
format Preprint
id arxiv_https___arxiv_org_abs_2601_23112
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle How Should AI Safety Benchmarks Benchmark Safety?
Yu, Cheng
Engelmann, Severin
Cao, Ruoxuan
Ali, Dalia
Papakyriakopoulos, Orestis
Computers and Society
AI safety benchmarks are pivotal for safety in advanced AI systems; however, they have significant technical, epistemic, and sociotechnical shortcomings. We present a review of 210 safety benchmarks that maps out common challenges in safety benchmarking, documenting failures and limitations by drawing from engineering sciences and long-established theories of risk and safety. We argue that adhering to established risk management principles, mapping the space of what can(not) be measured, developing robust probabilistic metrics, and efficiently deploying measurement theory to connect benchmarking objectives with the world can significantly improve the validity and usefulness of AI safety benchmarks. The review provides a roadmap on how to improve AI safety benchmarking, and we illustrate the effectiveness of these recommendations through quantitative and qualitative evaluation. We also introduce a checklist that can help researchers and practitioners develop robust and epistemologically sound safety benchmarks. This study advances the science of benchmarking and helps practitioners deploy AI systems more responsibly.
title How Should AI Safety Benchmarks Benchmark Safety?
topic Computers and Society
url https://arxiv.org/abs/2601.23112