Translated with AI from the Russian original.
AI scans database contents, recognizes personal data, and determines its categories
The company Gazinformservis has added an artificial intelligence-based tool for automatic search and classification of personal data to the domestic DBMS Jatoba. The new capability makes it possible to reduce database verification time and lower the risk of errors when preparing information for transfer to test environments or technical support.

The solution helps comply with the requirements of Law No. 152-FZ and find confidential information that must be concealed before data is used outside the core system. Unlike manual verification, the technology analyzes not only the database structure, but also the content of tables.
First, the system generates a list of tables and columns, then selects record samples from them. Regular expressions, dictionaries, and machine learning models are used for the search. Based on the analysis results, each column receives a classification: the system determines whether it contains personal data and assigns it to the corresponding category—for instance, full names or email addresses.
The search is linked to an anonymization tool. Based on the generated report and assigned security labels, the system can automatically mask the detected information. Dedicated algorithms are provided for different data categories, while administrators can configure their own processing rules.











