Jay Haybatov
February 6, 2024

Zero-knowledge security model: an introduction

Posted on February 6, 2024  •  5 minutes  • 939 words

News about ransom demands from cyber-criminals often features 7- and 8-digit numbers. Massive amounts of personal data and sensitive information from data leaks get sold or openly published on the internet. All of these can be stopped.

The zero-knowledge security model (ZKSM) is designed to guard sensitive and protected data even in cases of full access to the network infrastructure and/or cloud services. The ZKSM employs multiple techniques to prevent data theft by unauthorised parties:

ZKSM guarantees the following properties for data:

In realistic scenarios, the requirements beyond the ZKSM include:

Fully homomorphic encryption (FHE) – a technique that allows performing mathematical operations directly on encrypted data – can be optionally used for data processing. Modern performant FHE systems rely on approximate operations (e.g., get close but not exact results of multiplication), which limits their universal adoption. Despite this, 4^th^ generation FHE has many useful applications in secure data processing.

Often, companies advertise their end-to-end encrypted products as having ‘zero-knowledge’ properties. While end-to-end encryption satisfies some requirements of the ZKSM, it is only a part of the requirements. For example, all E2EE file sharing platforms, with a notable exception, use links for data sharing. This approach violates several ZKSM principles:

Before delving into the details of the ZKSM, the designer of a ZKSM system has a few considerations to make:

About the name

The zero-knowledge security model shares a part of its name with the zero-knowledge proof in cybersecurity. While these concepts are not directly related, they share the approach of proving that users have access to certain data without exposing any keys or passwords to the other party (server or cloud systems, in case of the ZKSM)

An application with end-to-end encryption alone does not qualify to be called a ‘zero-knowledge’ system as it lacks ways of proving users’ ownership of data/keys. For example, a few videoconferencing solutions advertise their end-to-end encryption capabilities despite ignoring the identity of parties joining the calls.

The zero-knowledge property must apply to the entire world – except the owner of data and her trusted parties. A system that advertises that it does not have access to data fulfils only a small fraction of ZKSM requirements. ‘Zero knowledge’ means that data is unreadable by anyone at storing, transmitting, sharing, and processing stages – unless authorised by the data owner.

to be continued…

Me somewhere else

There are other aspects to the life