On July 7, 2025, the European Commission and PIC/S released the latest revision of GMP, marking the first inclusion of artificial intelligence within the regulatory framework. The update sets clear boundaries: dynamic models, such as generative AI and large language models, cannot be directly applied in critical GMP contexts, while only static, deterministic models are permitted. Just as importantly, AI systems must remain verifiable, controllable, and ensure data integrity.

Visual inspection machines sit at the core of pharmaceutical quality control, and their performance directly determines both defect detection rates and regulatory compliance. Many industry players still rely on unsupervised learning methods, which compare products to a limited set of “normal” samples. However, these approaches often prove insufficient, as they often miss subtle or complex defects, produce high false positives due to batch variations, and lack the transparency required for GMP compliance.

Truking Technology takes a different approach by combining supervised learning with AIGC-based data enhancement.

• Supervised learning model: Mature frameworks for classification, segmentation, and object detection ensure each defect type is clearly labeled, delivering verifiable and controlled outputs fully aligned with GMP’s static model requirements.
• Comprehensive defect database: Built through years of industry experience, Truking’s dataset covers multiple vial types, materials, and defect categories, enabling robust training and continuous optimization.

• AIGC data augmentation: To address the scarcity of rare defect samples, Truking applies AI-generated content to produce realistic defect images. This expands the training set, enhances model generalization, and significantly improves robustness.

With this approach, Truking’s inspection machines achieve superior performance in detection accuracy, stability, and regulatory verifiability compared to unsupervised alternatives.

For customers, the benefits are clear:

• High-precision inspection of even the most complex or subtle defects.
• Regulatory compliance with GMP Chapter 4, Annex 11, and Annex 22, ensuring system validation and data integrity.
• Sustainable optimization through continuous data expansion and AI-based enhancements.

As AI becomes an integral part of GMP regulation, manufacturers face the challenge of balancing innovation with compliance. Truking Technology is already leading this transformation—delivering solutions that not only meet regulatory requirements but also advance pharmaceutical quality inspection into a smarter, more reliable future.