Syntax- and Compilation-Preserving Evasion of LLM Vulnerability Detectors
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911651175858176 |
|---|---|
| author | Sun, Luze Oprea, Alina Wong, Eric |
| author_facet | Sun, Luze Oprea, Alina Wong, Eric |
| contents | LLM-based vulnerability detectors are increasingly deployed in CI/CD security gating, yet their resilience to evasion under syntax- and compilation-preserving edits remains poorly understood. We evaluate five attack variants spanning four carrier families of behavior-preserving code transformations on a unified C/C++ benchmark ($N=5000$) and introduce Complete Resistance (CR), measuring the fraction of correctly detected vulnerabilities that withstand all attack variants. Our findings reveal a significant robustness gap: models achieving 70\%+ clean recall exhibit CR as low as 0.12\%, meaning over 87\% of detected vulnerabilities can be evaded by at least one syntax-preserving edit. Universal adversarial strings optimized on a 14B surrogate transfer effectively to black-box APIs including GPT-4o, while on-target optimization further amplifies evasion (up to 92.5\% ASR). These results indicate that clean benchmark accuracy alone is insufficient as a security guarantee for deployed vulnerability detectors. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_00305 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Syntax- and Compilation-Preserving Evasion of LLM Vulnerability Detectors Sun, Luze Oprea, Alina Wong, Eric Cryptography and Security Artificial Intelligence Machine Learning LLM-based vulnerability detectors are increasingly deployed in CI/CD security gating, yet their resilience to evasion under syntax- and compilation-preserving edits remains poorly understood. We evaluate five attack variants spanning four carrier families of behavior-preserving code transformations on a unified C/C++ benchmark ($N=5000$) and introduce Complete Resistance (CR), measuring the fraction of correctly detected vulnerabilities that withstand all attack variants. Our findings reveal a significant robustness gap: models achieving 70\%+ clean recall exhibit CR as low as 0.12\%, meaning over 87\% of detected vulnerabilities can be evaded by at least one syntax-preserving edit. Universal adversarial strings optimized on a 14B surrogate transfer effectively to black-box APIs including GPT-4o, while on-target optimization further amplifies evasion (up to 92.5\% ASR). These results indicate that clean benchmark accuracy alone is insufficient as a security guarantee for deployed vulnerability detectors. |
| title | Syntax- and Compilation-Preserving Evasion of LLM Vulnerability Detectors |
| topic | Cryptography and Security Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2602.00305 |