Metraipolis › archive › dc_cc1ff93205b4
Robustness and Trade-offs for Code LLMs on Protected Code — arXiv:2609.04220
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.04220
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T23:01:29.047Z
- SHA-256 of the archived text
7d1c8410e733f3608cecfa12ca27452c3c2afc0db722f3c86dcd416ac7128501- Stable URL of this page
https://metraipolis.com/documento/dc_cc1ff93205b4
How to cite
Robustness and Trade-offs for Code LLMs on Protected Code — arXiv:2609.04220 — arXiv (en), https://arxiv.org/abs/2609.04220 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cc1ff93205b4 (Metraipolis archive dc_cc1ff93205b4, sha256 7d1c8410e733f3608cecfa12ca27452c3c2afc0db722f3c86dcd416ac7128501)
BibTeX
@misc{dccc1ff93205b4,
title = {Robustness and Trade-offs for Code LLMs on Protected Code — arXiv:2609.04220},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.04220},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cc1ff93205b4; sha256 7d1c8410e733f3608cecfa12ca27452c3c2afc0db722f3c86dcd416ac7128501},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.04220. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Code large language models (LLMs) are increasingly used on software artifacts that may be intentionally obfuscated for intellectual-property protection, reverse-engineering resistance, or controlled access. In reverse engineering and security analysis, deobfuscation is commonly treated as the preprocessing step before downstream analysis or inference, yet its utility for code LLM pipelines has not been systematically validated across models and protection methods. We present an execution-based study of seven code LLMs on protected-code translation and completion, spanning source programs from C++, Go, Java, and JavaScript, five obfuscation methods, and three inference protocols: plain, obfuscated, and deobfuscated. Our results show that direct inference on obfuscated code often matches or exceeds inference on restored code. In this controlled benchmark, higher-capability models such as GPT-4.1 and Qwen3-Coder-30B retain about 90% Pass@1 on obfuscated translation inputs, indicating that explicit restoration is often unnecessary. Same-model restoration does recover some obfuscation-induced failures, but it also degrades many cases that already succeed, with lower-capability models showing the largest net losses. Across settings, model capability is the primary factor, while source language and obfuscation method have secondary but consistent effects. Overall, our findings support model-aware pipeline design and indicate that protected-code workflows should be evaluated primarily with execution-based metrics rather than static similarity alone. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-07 in cs.SE). Categorie arXiv: Software Engineering (cs.SE). Pagina del lavoro: https://arxiv.org/abs/2609.04220 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-16T22:37:16.592Z dalla pagina https://arxiv.org/catchup/cs.SE/2026-09-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).