Why it matters

Models, adapters and datasets are dependencies like code: a tampered or misrepresented artifact can bring backdoors, unsafe serialization or license problems into production.

How to prevent it

  • Obtain models and datasets from trusted sources, verify signatures or digests, and record their provenance.
  • Prefer safe weight formats such as safetensors to pickle-based files, and scan artifacts before loading them.
  • Keep an inventory of models and datasets, and evaluate third-party models before use.

Guides on this site

Main subject · Guide

Treat model releases as executable supply-chain inputs

Verify provenance, constrain loading and preserve the relationship between model weights, configuration and runtime dependencies.

References

  1. LLM03:2025 Supply Chain (OWASP, 2025 edition)
  2. OWASP Top 10 for LLM Applications 2026 (OWASP GenAI Security Project resource page)

List data from the publisher. A guide is listed as the main subject when the weakness or risk is what the guide is about, and as covering it when a dedicated section or worked example addresses it.