arXiv Artificial Intelligence

When LLM Decompilers Recompile More and Preserve Less

3 days ago 07.09.2026 07:00 👁 2 views
When LLM Decompilers Recompile More and Preserve Less

Quick summary

arXiv:2609.05370v1 Announce Type: cross Abstract: Decompilation recovers high-level source from compiled machine code and serves as a foundation for security tasks such as vulnerability detection and malware analysis. Traditional decompilers like Ghidra and Hex-Rays expose whatever they cannot resolve as visible placeholders and often emit pseudocode that will not compile or execute; LLM-based decompilers produce clean, idiomatic C and are now judged almost entirely by recompilability and re-executability: whether the output builds and passes its shipped input/output tests. We show that these

Key takeaways

  • arXiv:2609.05370v1 Announce Type: cross Abstract: Decompilation recovers high-level source from compiled machine code and serves as a foundation for security tasks such as vulnerability detection and malware analysis.
  • Traditional decompilers like Ghidra and Hex-Rays expose whatever they cannot resolve as visible placeholders and often emit pseudocode that will not compile or execute; LLM-based decompilers produce clean, idiomatic C and are now judged almost entirely by recompilability and re-executability: whether the output builds and passes its shipped input/output tests.

Why it matters

“When LLM Decompilers Recompile More and Preserve Less” shows why AI risk cannot be reduced to answer accuracy. Access controls, logging, human approval and incident response need to be designed into the workflow from the start.

Kaynak sitede devamını oku: arXiv Artificial Intelligence ↗