ImpossibleBench: Measuring LLMs' Propensity of Exploiting Test Cases

• URL: http://arxiv.org/abs/2510.20270v1
• Date: Thu, 23 Oct 2025 06:58:32 GMT
• Title: ImpossibleBench: Measuring LLMs' Propensity of Exploiting Test Cases
• Authors: Ziqian Zhong, Aditi Raghunathan, Nicholas Carlini,
• Abstract summary: "Shortcuts" to complete tasks pose significant risks for reliable assessment and deployment of large language models.<n>We introduce ImpossibleBench, a benchmark framework that measures LLM agents' propensity to exploit test cases.<n>As a practical framework, ImpossibleBench is not just an evaluation but a versatile tool.
• Score: 58.411135609139855
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The tendency to find and exploit "shortcuts" to complete tasks poses significant risks for reliable assessment and deployment of large language models (LLMs). For example, an LLM agent with access to unit tests may delete failing tests rather than fix the underlying bug. Such behavior undermines both the validity of benchmark results and the reliability of real-world LLM coding assistant deployments. To quantify, study, and mitigate such behavior, we introduce ImpossibleBench, a benchmark framework that systematically measures LLM agents' propensity to exploit test cases. ImpossibleBench creates "impossible" variants of tasks from existing benchmarks like LiveCodeBench and SWE-bench by introducing direct conflicts between the natural-language specification and the unit tests. We measure an agent's "cheating rate" as its pass rate on these impossible tasks, where any pass necessarily implies a specification-violating shortcut. As a practical framework, ImpossibleBench is not just an evaluation but a versatile tool. We demonstrate its utility for: (1) studying model behaviors, revealing more fine-grained details of cheating behaviors from simple test modification to complex operator overloading; (2) context engineering, showing how prompt, test access and feedback loop affect cheating rates; and (3) developing monitoring tools, providing a testbed with verified deceptive solutions. We hope ImpossibleBench serves as a useful framework for building more robust and reliable LLM systems. Our implementation can be found at https://github.com/safety-research/impossiblebench.

Related papers

• RealSec-bench: A Benchmark for Evaluating Secure Code Generation in Real-World Repositories [58.32028251925354]
  Large Language Models (LLMs) have demonstrated remarkable capabilities in code generation, but their proficiency in producing secure code remains a critical, under-explored area.<n>We introduce RealSec-bench, a new benchmark for secure code generation meticulously constructed from real-world, high-risk Java repositories.
  arXiv  Detail & Related papers  (2026-01-30T08:29:01Z)
• Automated structural testing of LLM-based agents: methods, framework, and case studies [0.05254956925594667]
  LLM-based agents are rapidly being adopted across diverse domains.<n>Current testing approaches focus on acceptance-level evaluation from the user's perspective.<n>We present methods to enable structural testing of LLM-based agents.
  arXiv  Detail & Related papers  (2026-01-25T11:52:30Z)
• CompassVerifier: A Unified and Robust Verifier for LLMs Evaluation and Outcome Reward [50.97588334916863]
  We develop CompassVerifier, an accurate and robust lightweight verifier model for evaluation and outcome reward.<n>It demonstrates multi-domain competency spanning math, knowledge, and diverse reasoning tasks, with the capability to process various answer types.<n>We introduce VerifierBench benchmark comprising model outputs collected from multiple data sources, augmented through manual analysis of metaerror patterns to enhance CompassVerifier.
  arXiv  Detail & Related papers  (2025-08-05T17:55:24Z)
• Do AI models help produce verified bug fixes? [62.985237003585674]
  Large Language Models are used to produce corrections to software bugs.<n>This paper investigates how programmers use Large Language Models to complement their own skills.<n>The results are a first step towards a proper role for AI and LLMs in providing guaranteed-correct fixes to program bugs.
  arXiv  Detail & Related papers  (2025-07-21T17:30:16Z)
• The SWE-Bench Illusion: When State-of-the-Art LLMs Remember Instead of Reason [1.6249398255272318]
  We present empirical evidence that performance gains on SWE-Bench-Verified may be partially driven by memorization rather than genuine problem-solving.<n>We show that state-of-the-art models achieve up to 76% accuracy in identifying buggy file paths using only issue descriptions, without access to repository structure.<n>These findings raise concerns about the validity of existing results and underscore the need for more robust, contamination-resistant benchmarks.
  arXiv  Detail & Related papers  (2025-06-14T00:25:26Z)
• OSS-Bench: Benchmark Generator for Coding LLMs [4.393587297483245]
  We introduce OSS-Bench, a benchmark generator that constructs large-scale, live evaluation tasks from real-world open-source software.<n> OSS-Bench replaces functions with LLM-generated code and evaluates them using three natural metrics: compilability, functional correctness, and memory safety.<n>Our results demonstrate that OSS-Bench mitigates overfitting by leveraging the evolving complexity of OSS.
  arXiv  Detail & Related papers  (2025-05-18T09:53:51Z)
• Towards Automated Formal Verification of Backend Systems with LLMs [9.66648456498893]
  We propose a novel framework that leverages functional programming and type systems to translate backend code into formal Lean representations.<n>Our pipeline automatically generates theorems that specify the intended behavior of APIs and database operations, and uses LLM-based provers to verify them.<n>We evaluate our method on realistic backend systems and find that it can formally verify over 50% of the test requirements, which suggests that half of a testing engineer's workload can be automated.
  arXiv  Detail & Related papers  (2025-04-13T16:49:37Z)
• Learning to Ask: When LLM Agents Meet Unclear Instruction [55.65312637965779]
  Large language models (LLMs) can leverage external tools for addressing a range of tasks unattainable through language skills alone.<n>We evaluate the performance of LLMs tool-use under imperfect instructions, analyze the error patterns, and build a challenging tool-use benchmark called Noisy ToolBench.<n>We propose a novel framework, Ask-when-Needed (AwN), which prompts LLMs to ask questions to users whenever they encounter obstacles due to unclear instructions.
  arXiv  Detail & Related papers  (2024-08-31T23:06:12Z)
• LiveBench: A Challenging, Contamination-Limited LLM Benchmark [93.57775429120488]
  We release LiveBench, the first benchmark that contains frequently-updated questions from recent information sources.<n>We evaluate many prominent closed-source models, as well as dozens of open-source models ranging from 0.5B to 405B in size.<n>Questions are added and updated on a monthly basis, and we release new tasks and harder versions of tasks over time.
  arXiv  Detail & Related papers  (2024-06-27T16:47:42Z)
• AgentQuest: A Modular Benchmark Framework to Measure Progress and Improve LLM Agents [19.439775106707344]
  AgentQuest is a framework where benchmarks and metrics are modular and easily through well documented and easy-to-use APIs. We offer two new evaluation metrics that can reliably track LLM agent progress while solving a task. We exemplify the utility of the metrics on two use cases wherein we identify common failure points and refine the agent architecture to obtain a significant performance increase.
  arXiv  Detail & Related papers  (2024-04-09T16:01:24Z)
• ML-Bench: Evaluating Large Language Models and Agents for Machine Learning Tasks on Repository-Level Code [76.84199699772903]
  ML-Bench is a benchmark rooted in real-world programming applications that leverage existing code repositories to perform tasks. To evaluate both Large Language Models (LLMs) and AI agents, two setups are employed: ML-LLM-Bench for assessing LLMs' text-to-code conversion within a predefined deployment environment, and ML-Agent-Bench for testing autonomous agents in an end-to-end task execution within a Linux sandbox environment.
  arXiv  Detail & Related papers  (2023-11-16T12:03:21Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.