Real-time prevention of misappropriation and/or exposure assessment posed by a transaction at a POS terminal, especially those transactions in which it is not possible to otherwise definitively determine that a user/transactor is present at the POS terminal. AI is relied on, such as, a neural network implementing sequence modeling, to predict the likelihood that a user/transactor is present at the POS terminal. Location of a user is tracked via a designated mobile communication device and in instances in which a location of the mobile communication cannot be identified at the time of the transaction, the invention performs analysis based at least on the last-in-time location of the mobile communication device, the user's travel route leading up to the last-in-time location of the mobile communication device and other factors to predict a likelihood that the user is located at the POS terminal. As a result of the predicted likelihood either the transaction is automatically decisioned or the user/transactor is alerted and requested to their approve or reject/deny the transaction.