A large volume of unstructured data exists in the automotive industry and needs to be analyzed to detect potential vehicle concerns. Much of this data is textual in nature since customer complaints are made through call center interactions and warranty repairs. Current approaches to detect potential vehicle concerns in text data include various keyword search methods. In this paper, we apply Natural Language Processing (NLP) and shallow machine learning methods on text data to create classifiers to detect the potential vehicle concern of airbag non-deployment. For this potential vehicle concern, we show the performance of multinomial Naïve Bayes (NB), Support Vector Machine (SVM) and Gradient Boosted Trees (GBT) classifiers against keyword search methods. We present challenges of classification model development related to the nature of automotive data and limited training data. Our findings provide insights on robust text classification approaches that can improve identification of potential vehicle concerns.