<p>The authentication of extra virgin olive oil (EVOO) is an essential aspect of food quality control due to its high commercial value and susceptibility to economically motivated adulteration. This study presented a comprehensive analytical strategy combining gas chromatography-mass spectrometry (GC-MS) with machine learning (ML) models to detect and classify EVOO adulteration. Chemical profiling focused on squalene and phytosterols, and notably, six phytosterols were identified in EVOO for the first time, enriching its compositional fingerprint. A dataset was constructed using pure EVOO, common edible oils (soybean, rapeseed, peanut, and corn oils), and mixtures with EVOO at adulteration levels ranging from 1 to 40%. Four classification algorithms were evaluated: partial least squares discriminant analysis (PLS-DA), orthogonal PLS-DA (OPLS-DA), decision tree (DT), and k-nearest neighbors (KNN). Classical chemometric models (PLS-DA and OPLS-DA) achieved moderate external validation accuracy (82.1%). In contrast, the KNN model demonstrated excellent performance, with 100% accuracy in the test set and 94.7% in the validation set. Importantly, it could detect adulteration at levels as low as 1%, indicating high sensitivity and robustness. These results underscore the practical utility of integrating GC-MS with interpretable ML models, especially KNN, for sensitive and reliable identification of adulteration in high-value edible oils. The proposed approach holds significant promise for applications in food authentication, quality assurance, and regulatory monitoring.</p>