Do Androids Dream of Breaking the Game? Systematically Auditing AI Agent Benchmarks with BenchJack
New arXiv research introduces BenchJack, an automated red-teaming system to audit AI agent benchmarks for reward hacking. The authors derive a taxonomy of eight flaw patterns from past incidents and apply BenchJack to 10 popular benchmarks, discovering 219 distinct flaws. An iterative generative-adversarial pipeline reduces hackable-task ratio from near 100% to under 10% within three iterations, fully patching WebArena and OSWorld. Results show evaluation pipelines lack adversarial mindset, and proactive auditing can close the security gap.
[2605.12673] Do Androids Dream of Breaking the Game? Systematically Auditing AI Agent Benchmarks with BenchJack
[Submitted on 12 May 2026]
Title:Do Androids Dream of Breaking the Game? Systematically Auditing AI Agent Benchmarks with BenchJack
View a PDF of the paper titled Do Androids Dream of Breaking the Game? Systematically Auditing AI Agent Benchmarks with BenchJack, by Hao Wang and 5 other authors
View PDF HTML (experimental)
Abstract:Agent benchmarks have become the de facto measure of frontier AI competence, guiding model selection, investment, and deployment. However, reward hacking, where agents maximize a score without performing the intended task, emerges spontaneously in frontier models without overfitting. We argue that benchmarks must be secure by design. From past incidents of reward hacks, we derive a taxonomy of eight recurring flaw patterns and compile them into the Agent-Eval Checklist for benchmark designers. We condense the insights into BenchJack, an automated red-teaming system that drives coding agents to audit benchmarks and identify possible reward-hacking exploits in a clairvoyant manner. Moreover, we extend BenchJack to an iterative generative-adversarial pipeline that discovers new flaws and patches them iteratively to improve benchmark robustness. We apply BenchJack to 10 popular agent benchmarks spanning software engineering, web navigation, desktop computing, and terminal operations. BenchJack synthesizes reward-hacking exploits that achieve near-perfect scores on most of the benchmarks without solving a single task, surfacing 219 distinct flaws across the eight classes. Moreover, BenchJack's extended pipeline reduces the hackable-task ratio from near 100% to under 10% on four benchmarks without fatal design flaws, fully patching WebArena and OSWorld within three iterations. Our results show that evaluation pipelines have not internalized an adversarial mindset, and that proactive auditing could help close the security gap for the fast-paced benchmarking space.
Subjects:
Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
Cite as: arXiv:2605.12673 [cs.AI]
(or arXiv:2605.12673v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2605.12673
arXiv-issued DOI via DataCite (pending registration)
Submission history
From: Hao Wang [view email] [v1] Tue, 12 May 2026 19:22:45 UTC (1,741 KB)
Full-text links:
Access Paper:
View a PDF of the paper titled Do Androids Dream of Breaking the Game? Systematically Auditing AI Agent Benchmarks with BenchJack, by Hao Wang and 5 other authors
View PDF
HTML (experimental)
TeX Source
view license
Current browse context:
cs.AI
new | recent | 2026-05
Change to browse by:
cs cs.CR
References & Citations
NASA ADS
Google Scholar
Semantic Scholar
Loading...
Data provided by:
Bibliographic Tools
Bibliographic and Citation Tools
Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
Connected Papers Toggle
Connected Papers (What is Connected Papers?)
Litmaps Toggle
Litmaps (What is Litmaps?)
scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
Code, Data, Media
Code, Data and Media Associated with this Article
alphaXiv Toggle
alphaXiv (What is alphaXiv?)
Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub Toggle
DagsHub (What is DagsHub?)
GotitPub Toggle
Gotit.pub (What is GotitPub?)
Huggingface Toggle
Hugging Face (What is Huggingface?)
ScienceCast Toggle
ScienceCast (What is ScienceCast?)
Demos
Demos
Replicate Toggle
Replicate (What is Replicate?)
Spaces Toggle
Hugging Face Spaces (What is Spaces?)
Spaces Toggle
TXYZ.AI (What is TXYZ.AI?)
Related Papers
Recommenders and Search Tools
Link to Influence Flower
Influence Flower (What are Influence Flowers?)
Core recommender toggle
CORE Recommender (What is CORE?)
Author
Venue
Institution
Topic
About arXivLabs
arXivLabs: experimental projects with community collaborators
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.
Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)