Research question
Why can a pipeline remain structurally valid while its business meaning becomes wrong?
Core idea
Generative components can introduce outputs that satisfy schema and type checks while changing the intended meaning. The failure may propagate because downstream systems see valid structure, not semantic corruption.
Why it matters
Enterprise controls often assume malformed data will reveal itself. AI-assisted transformations weaken that assumption. Detection has to include provenance, context, and semantic checks.
Contribution
The work distinguishes structural validity from semantic correctness and frames provenance, context, and meaning-aware controls as necessary for GenAI-augmented data pipelines.
Limitations
The framework does not imply that semantic correctness can be fully automated. Domain ownership, review, and context remain necessary when meaning is ambiguous.
Status
Accepted for the IEEE ICDCS 2026 Industry Program.