JsDeObsBench : Measuring and Benchmarking LLMs for JavaScript Deobfuscation Conference: Proceedings of the 2025 ACM SIGSAC Conference on Computer and Communications Security; October 13–17, 2025; Taipei, Taiwan. Proceedings of the 2025 ACM SIGSAC Conference on Computer and Communications Security (C
Guoqiang Chen, Xin Jin, Zhiqiang Lin · 2025
Read JsDeObsBench : Measuring and Benchmarking LLMs for JavaScript Deobfuscation Conference: Proceedings of the 2025 ACM SIGSAC Conference on Computer and Communications Security; October 13–17, 2025; Taipei, Taiwan. Proceedings of the 2025 ACM SIGSAC Conference on Computer and Communications Security (C by Guoqiang Chen, Xin Jin, Zhiqiang Lin (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-20170.txt.
Description
Chen, Jin, and Lin built a benchmark in twenty twenty-five for testing how well large language models clean up scrambled JavaScript code. Obfuscation deliberately garbles code, a trick often used to hide malware. Automating its reversal could aid web security, but no systematic test existed.
Their benchmark, JsDeObsBench, spans tricks from simple variable renaming to deep structural rewrites. Leading models, including GPT-4o and several open code models, were tested against traditional tools. The models simplified code better than the baselines but stumbled more on exact syntax and on keeping the code runnable. Trials on real JavaScript malware showed practical promise. The authors map where models excel and where they still fail, to guide safer deobfuscation tools.
Catalog record
| CREATOR | Guoqiang Chen, Xin Jin, Zhiqiang Lin |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 99 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-20170