Researchers are evaluating Large Language Models (LLMs) to assist in identifying the processor architecture of bare-metal binaries. One strategy involves using Ghidra to disassemble the binary across numerous architectures and then employing LLMs to analyze the disassembly outputs. Several LLMs, including Qwen2.5-coder, Qwen3-coder:30b, dolphinMistral24b, and dolphin3-cyber, were tested on different firmware types, with varying success in identifying candidate processors and generating consistent reports. AI
IMPACT This research could improve the efficiency and accuracy of reverse engineering efforts by automating processor identification.
RANK_REASON The item details a research methodology and benchmark results for using LLMs in reverse engineering tasks. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
