Refrigeration Cyber Incidents as a Warning for AI-Enabled Maintenance: Evidence Review and Security Requirements

This preprint examines publicly reported cyber incidents involving commercial refrigeration controllers, with particular attention to the October and November 2026 incidents in Europe and related historical and United States accounts. It reviews six public-source incident clusters and distinguishes official confirmation, attributed retailer or vendor accounts, suspected causation, prosecution allegations, demonstrated technical capability and unresolved inference. The paper proposes a five-stage framework connecting digital access, control manipulation, physical-process response, monitoring evidence and recovery decisions. It then applies six security requirements to prospective AI-enabled maintenance workflows where operational reports or other lower-trust content may influence equipment commands. The analysis does not claim that AI caused the reviewed incidents and does not report a new exploit, model evaluation or thermal experiment. The accompanying evidence artifact contains machine-readable incident and claim registers, the supporting schema, documentation and citation metadata. These files are provided to make the evidence classifications, wording limits and analytical boundaries easier to inspect and reproduce.