AI becomes a documented preliminary finding in official meat inspection

Official meat inspection is entering a phase in which machine-vision systems can provide systematic preliminary findings. At Westfleisch’s Oer-Erkenschwick plant, the “ArtificialVet” pilot project has used six high-performance cameras since December 2025 to capture the closed pig carcass. The system supplies digital preliminary findings to official inspection personnel; it does not replace the veterinary decision. The technical significance lies in the sequence of work. Such a system can structure observation and direct attention to visible abnormalities. That can make documentation more consistent and help manage the large number of carcasses. Yet the quality of a preliminary finding depends on defined imaging conditions, suitable training data and clear processes for checking divergent results. Cameras do not make everything visible: assessments involving organs, pathogens, or matters requiring tactile or laboratory examination still require complementary procedures. For plants, the pilot moves the AI debate from abstract promises to concrete integration into regulated processes. The key question is not whether an algorithm identifies an abnormality in an image, but how that finding is documented, reviewed and incorporated into the official decision. This requires unambiguous roles for operator, authority and inspector, as well as traceability of images, model versions and corrections. The pilot merits attention precisely because of this boundary. If digital preliminary findings prove reliable in practice, the technology can strengthen risk-based inspection without diluting professional accountability. For the industry, the value lies less in spectacular automation than in a more robust, auditable chain of evidence.