Skip to the content
AI and Cybersecurity Learning and Practice Center Test automation Open the Learning Center

WeaknessASC-2026-0060Other weakness

Unverified model provenance

MediumFixed

Models are downloaded by name from public hubs without checking who published them, whether the repository was renamed or hijacked, or whether the files changed.

How to detect it

  • Models referenced by name without revision or hash
  • Models from accounts that recently changed names or owners

How to resolve it

  • Pin models to a specific revision and verify hashes
  • Mirror approved models to an internal registry

How to investigate

  1. Compare the deployed model's revision and hash with the publisher's
  2. Replace unverified models

Sources

Open it in the Learning Center's vulnerability database