AI-Driven Image Analysis for Enhancing Audits in Food and Beverage Industries Under GMP and SSOP Standards

Abstract

This study evaluated the application of artificial intelligence (AI)-driven image analysis for enhancing Good Manufacturing Practices (GMP) and Sanitation Standard Operating Procedures (SSOP) audits in food and beverage (F&B) catering services. Three industrial F&B establishments located in Jakarta, Cikarang, and Karawang were assessed. Images capturing key visual criteria were analyzed using an AI system based on ChatGPT-4o, with compliance scored under three different prompt formulations to evaluate AI sensitivity. Manual audits conducted by trained auditors served as a benchmark. Statistical analysis revealed that AI assessments closely aligned with manual audits across most criteria, particularly for cleanliness of food-contact surfaces, personal hygiene, and pest exclusion. However, significant prompt-induced differences were found in more interpretative criteria such as facility design and storage practices. When averaged across stable prompts, AI scores showed strong agreement with manual audits, although AI tended to assign slightly stricter scores in certain areas. No significant differences were found in SSOP compliance evaluations, indicating high consistency for sanitation-related assessments. These results demonstrate that AI-driven image analysis can reliably support GMP and SSOP audits for visually detectable parameters, improving audit efficiency, objectivity, and frequency. Nonetheless, non-visual aspects such as documentation and microbiological testing still require human oversight. Integrating AI into food safety auditing represents a promising advancement for modern F&B compliance monitoring.