We're the leading agency in artificial intelligence & machine learning

   +31 6 53 64 34 73   St. Jacobsstraat 12, Utrecht

International Bank

Leaked Secret Detection in Documents and Corporate Wikis

Real-time warning

Classification

Challenge

Warn when sensitive information, such as passwords, is shared.

Target Group

Administrators of the knowledge management system.

Results

A smart system that detects sensitive information, making the knowledge management system more secure.

Collaboration and sharing information

Companies are increasingly collaborating to launch new products, services, or innovations. For this, employees of different companies must be able to share information confidentially. Because it sometimes concerns competition or stock market-sensitive information, confidentiality is very high.

We created a platform where corporate requirements such as onboarding, and other compliance requirements are guaranteed to keep this safe and efficient. This platform is used by international central banks, among others.

Challenge

Because large groups of employees from different companies work together and use the collaboration platform, authorization and control are essential. Although good agreements are in place, users can still accidentally share sensitive information. Because the system has many users who add or change information daily, it becomes increasingly challenging to detect sensitive information.

Solution

All user additions and changes route through a funnel of machine learning models, which checks the data for sensitivities. First, there’s a coarse search for sensitive words, such as passwords, then words are examined in the context of the sentence, and at the bottom of the funnel, a classification is made based on allowed patterns. A phone number can look like a password, but of course, it isn’t.

The system reports the outcome to the administrators of the system. They will receive a notification containing a reference to the document where sensitive information is shared. The administrators make the final judgment about the sensitivity of the data.

Impact

The platform is gradually improving without any inconvenience to users. Controlling sensitive information is getting smarter and alerting administrators so they can act. It makes sharing information more secure.

 

Got interested? Get in touch.

Accelerator

Use Aigeny's accelerator to speed up development

Setting up an MLOps platform & pipeline in the cloud to continuous deliver Machine Learning Models.

Accelerator

Use Aigency's accelerator to speed up

Get a proof-of-value of a Machine Learning Model within 5 days with our standard approach.