A security team is reviewing an AI model that was fine-tuned using a third-party dataset downloaded from a public repository. Before approving production deployment, which supply chain security check is most critical?

Question

Accepted Answer

A. Verify the cryptographic hash of the dataset against the publisher's published checksum and review the dataset's provenance and license to confirm it has not been tampered with or contains injected malicious samples. Cryptographic hash verification against the publisher's checksum detects tampering; provenance review confirms the dataset comes from a trustworthy source and identifies potentially injected malicious samples before they corrupt the model. The SecAI+ objectives include supply chain security for AI training data. File size checks are insufficient as malicious modifications can be size-neutral. HTTPS ensures transport security but does not verify the dataset's content integrity or provenance.

A security team is reviewing an AI model that was fine-tuned using a third-party dataset downloaded from a public repository. Before approving production deployment, which supply chain security check is most critical?

Related Questions

Discussion