The Russian company FIS, a Skolkovo resident, has developed a service that allows banks and other organizations to work with external neural networks without transferring original confidential data to them. The program automatically hides personal and corporate information in documents while retaining the data necessary for artificial intelligence to analyze.

Russian service will allow banks to use external AI without disclosing confidential data
Image source: ChatGPT

The technology works as follows: before sending a document to the neural network, the local system finds names, company names, bank details, and other sensitive information in it. Conditional designations are substituted instead. For example, AI will be able to analyze a contract between two organizations, compare terms, and find contradictions, but will not receive their real names and details.

The main feature of the development is the ability to restore hidden information after processing. To do this, the program saves a special correspondence table, which remains within the company's protected infrastructure. If the data is not needed later, it can be deleted without the possibility of recovery.

The service supports PDF, Word, Excel, PowerPoint, and other formats, including diagrams and images with text. It can be used via a web interface or integrated into corporate systems. Deployment is provided both on the organization's own servers and in a secure cloud.

FIS expects that the technology will help companies leverage the capabilities of powerful external AI models without the need to create their own expensive computing systems. At the same time, the actual security of processing will depend on the correctness of anonymization and compliance with internal information protection requirements.

Read more on the topic:

Never miss our newsAdd this site to your preferred sources to see us more oftenAdd on Google

Сейчас на главной

Все новости