In order to prevent sensitive information leakage, sometimes called data leakage or loss to unauthorized recipients, organizations use information security systems to prevent it. In fact, data can leak through a variety of channels. While it is impossible to completely prevent it from happening, there are certain things one can do to decrease the likelihood of it happening. All financial institutions manage client information for commercial purposes. Humans used to search these digital documents for sensitive information, which was labor-intensive and expensive. In order to analyze material using cutting-edge data mining, statistics, and machine learning methods across multiple data dimensions, a smart and trustworthy system is required. According to the research, AI-based information leak prevention solutions that require no configuration can be created by utilizing LSTM to categorize document pictures according to the existence of NPI and PII semantic signatures. This system is made to be actively utilized as a tag SD pictures when they are storing data alarm systems. It may also serve as a real-time checkpoint for data loss brought on by documents that are being utilized or transported. The proposed model offers a protection mechanism against information loss by utilizing the most sophisticated LSTM-based binary classifier in artificial intelligence.