Early data leak detection - with AI against internal data theft
The Challenge
Data leaks can be costly for companies. Unintentional access to customer data, for example, can quickly result in horrendous fines—keyword: data protection. Not to mention the significant loss of image and trust in the public eye. Numerous examples from the recent past show how important it is for companies to protect themselves against data leaks. Internal data leaks can occur, for example, as a result of industrial espionage, internal data theft, or hacked employee accounts. According to Statista, a data leak costs companies in Germany an average of 4.5 million euros. It takes around 160 days to close the leak. Early detection speeds up the process and reduces costs – ideally, data theft can be avoided altogether. We developed an early detection system for data leaks for an international mobile phone provider that uses AI to reliably flag unusual behavior in SQL queries. The goal was to detect potential unwanted data access in time to take quick countermeasures.
Our solution
The development team - consisting of data engineers and data scientists - opted for a machine learning approach in which the statistical distribution of SQL activities on the data warehouse is analyzed and deviating behaviour is displayed. The data analysis and the AI algorithm form the basis of the early warning system. Pre-processing of all data was necessary for the modeling - in this case over 1.5 billion data records with a data volume of around 3 terabytes. In order to be able to process this large amount of information, a powerful cloud platform was implemented with the Data Platform. The SQL activities were then classified, the normal behavior analyzed and various criteria for the evaluation determined. The processed data was used to train the AI. Deviations from normal behavior are displayed and evaluated according to a predefined points system.
The result
The data leak early detection system was explicitly adapted to the customer's needs and wishes. Conspicuous behavior in SQL data queries is displayed in good time so that our customer can reliably check whether it is an internal data leak. In addition to improved protection against data theft, the solution offers increased transparency in database activities and IT management can be optimized accordingly. The program code has become the property of the customer so that internal IT specialists can use it freely.
The most important facts in brief
Customer benefits
- Detection and prevention of data theft and industrial espionage
- Avoidance of loss of image and trust
- Avoidance of fines and loss of sales
- Optimization of existing processes
- Transparency in data movements
- Strengthening employees' skills in handling sensitive data
The details
Who is it suitable for? For all companies that use sensitive data in their databases and have their own IT department.
_hyand_success_stories
Discover our digital solutions
You too can benefit from our extensive experience with projects involving companies of all sizes.
HyExpert - the IT talk.
Would you like to know how you can take your company forward with our customized IT solutions? Do you have questions about specific digital challenges?
Contact us - we'll help you!