From transcript to trusted portfolio: A provenance-centered model for AI and cybersecurity transfer pathways

Higher education still lacks reliable ways to document the authentic provenance of student computational work. Transcripts record course numbers and grades, but do not preserve a dataset a student cleaned, a container within which a model was trained, a random seed, a threat model behind a security analysis, or a keystroke trail that establishes authorship. This paper proposes a design science, work-in-progress method for producing a provenance-centered portfolio for artificial intelligence and cybersecurity education pathways. The provenance portfolio provides value in student transfer contexts where receiving institutions must read across heterogeneous curricula. Drawing on the accountability logic of AS9100 aerospace quality systems, the model fuses cryptographic authorship verification, W3C PROV lineage metadata, Verifiable Credentials, and 1EdTech Open Badges 3.0 into a portable, machine-checkable evidence layer. We specify a five-layer cyberinfrastructure architecture, an eight-vector threat taxonomy with a novel selective-scrutiny attack we name T8, and an automation-first verification pipeline whose pass/fail authenticity check is deterministic, leaving qualitative judgment to human faculty. A sample lightweight Python implementation (47 KB, 1,200 LOC, RSA-PSS signing, PROV-JSON output) is released and exercised end-to-end against a demo PyTorch artifact, producing a verification report scoring 1.0 on provenance completeness. Mapping artifacts to NSA/DHS CAE knowledge units (Becker, A., et al., 2024) transforms pedagogy by treating reproducibility as a Week-1 competency. The model addresses the converging problem of distinguishing student-authored from AI-assisted (or AI-produced) work while making transfer credentials portable, reproducible, and verifiable.