The Lot That Could Not Be Found · Food traceability, uncertain inventory and the evidentiary limits of artificial intelligence after the FDA's November 2026 consultation
The United States Food and Drug Administration's November 2026 consultation on lot-level traceability exposes a fundamental problem in contemporary food logistics: an inventory system may know how many cases remain without knowing which production lots they represent. The distinction matters whenever products are commingled, repacked, transformed or moved through facilities that record transactions more reliably than physical identity. This article analyses the legal architecture of the US Food Traceability Rule, its deferred enforcement, and the different obligations operating in the European Union, Spain and Great Britain. It then develops a technical framework for distinguishing observed lot identity from inferred provenance, quantifying uncertainty, validating algorithms and testing recall performance. Particular attention is given to mixed pallets, continuous production, ingredient transformation and microbiological incident response. The central proposition is that digital traceability should be evaluated as an evidentiary system, not merely as a database or software procurement exercise. A proposed research agenda examines whether probabilistic reconstruction can reduce investigation time without creating unjustified certainty or narrowing recalls unsafely.
Authors
- Jose Manuel López (ORCID: https://orcid.org/0009-0006-9238-0080)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-15
- DOI
- https://doi.org/10.5281/zenodo.24244859
- Primary Topic
- Food Supply Chain Traceability
- Type
- article
- Field-Weighted Citation Impact
- 0.00