2026 · Preprint
SoK: AI-Augmented Binary Reversing
A systematization-of-knowledge survey of machine-learning, LLM and agentic approaches to binary reversing. Version 2 collects 246 papers published since 2015, found through keyword searches of 11 security, software-engineering and AI venues plus citation snowballing, and organizes them into 22 domains by inference task: from code-boundary detection and recovery of function names, types and struct layouts to code similarity, malware and vulnerability analysis, and decompilation. Its taxonomy links conventional triage, static, dynamic and security-testing analyses to AI pipelines through the binary-derived artifacts they exchange, and it derives 12 insights and open directions such as binary-aligned ground truth and goal-driven agentic reversing.
Scope and limitations
A literature systematization with no new experiments, benchmark or tool; it classifies and summarizes reported results without reproducing them. The corpus covers papers available up to July 31, 2026, applies citation thresholds that fall from 12 for 2015–2016 papers to 2 for 2025–2026 papers, excludes methods that need source code at inference time, and leaves out Android and interpreted-language bytecode. Detailed per-stage tables draw on 88 top-venue studies rather than the full corpus. It is an arXiv preprint with no stated venue, and this record makes no claim of peer review. Papers it cites are not thereby part of this collection; each needs its own review.
Why it fits this collection
Provides a structured map of ML, LLM and agent research across binary reversing tasks, helping place the collection’s binary-analysis papers and locate related work that still needs individual review.
Publication and review details
- Venue
- arXiv
- Status
- Preprint
- Publication / posting date
- June 16, 2026
- Date source
- arXiv v1 submission; the reviewed v2 revision of 2026-09-04 does not change the first-submission date.
- First arXiv submission
- June 16, 2026
- arXiv
- 2606.17398
- Reviewed arXiv version
- v2
Reviewed arXiv v2 (2026-09-04; comment “21 pages, 7 tables, 4 figures”), which lists five authors and 246 papers; this record follows the v2 author list. v1 (2026-06-16; “20 pages, 7 tables, 3 figures”) listed six authors, Yujeong Kwon, Yiyue Zhang, Shakhzod Yuldoshkhujaev, Kexin Pei, Dokyung Song and Hyungjoon Koo, and analyzed 144 papers published between 2015 and 2025. Neither version states a venue or journal reference. Footnote 1 in both versions links the companion site, an interactive JavaScript table of the surveyed papers grouped by domain with links to their publications; no source repository is linked, so code_url is null. The site and its paper list may change after this review. No experiments or site code were run.
Sources reviewed
Source review only; experiments were not rerun for this collection. Cite the original authors using the linked paper or BibTeX.