Reproduce “Forensic Implications of Localized AI: Artifact Analysis of Ollama, LM Studio, and llama.cpp” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Shariq Murtuza

This research systematically investigates the digital forensic artifacts of popular local Large Language Model runners, Ollama, LM Studio, and llama.cpp, on Windows and Linux operating systems. It documents their unique evidentiary footprints, demonstrating the recoverability of user interactions and underscoring the necessity of memory forensics for certain ephemeral LLM activities.

Key insights

  • The architectural design of local LLM runners (client-server, Electron app, CLI) directly dictates the nature, persistence, and recoverability of digital artifacts.
  • LM Studio generates the most structured and comprehensive forensic artifacts, including persistent chat histories and RAG pipeline caches.
  • llama.cpp, due to its minimalist design, produces the most ephemeral artifacts, making memory forensics critical for reconstructing interactive user prompts.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.