Know Center Research

Data Privacy for AI

How can we make machine learning and AI secure enough to be accepted by broad segments of the population? Most importantly, how can we use cryptographic methods today that we know will still be secure in 20 years, no matter how fast the quantum computer trying to decrypt them will be?

Data Privacy for AI

Research Focus

In the research area “Data Privacy for AI”, we research and develop efficient and long-term secure (quantum computer safe) cryptographic methods that can be used for a wide range of practical applications. We create secure foundations for trustworthy AI and also overcome the performance problems that still exist today with complex machine learning algorithms.
Longterm-secure cryptographic methods

We are working on the development of new cryptographic methods that have already convincingly shown that they are also resistant to the threat of quantum computing and can be used as privacy-enhancing technologies (PETs). Concrete further goals are the design and security analysis of new secure post-quantum methods that are secure against both classical and quantum attacks.

Privacy-Preserving Data Analytics

We aim to advance the state-of-the-art for privacy-compatible methods such as HE, MPC, or PSI and overcome performance issues typically encountered in more complex evaluations such as machine learning applications. We are developing long-term secure cryptographic systems that can be used for privacy-compatible data analysis, especially in the context of machine learning and AI.

Quantum Information

We bring together our expertise in cryptography and analytics with quantum information and quantum algorithms. Given recent breakthroughs in the development of quantum computers, the scenario of quantum computing is becoming increasingly realistic. However, because users of quantum computers have to share their data with operators, many questions about privacy and data security are raised.

Potential

Protect your competitive advantages!

With the use of AI, it is possible to derive entirely new correlations and models from large amounts of data, which decisively advance companies, science, and society. However, protecting data from misuse and ensuring confidentiality have top priority. This is a prerequisite for handling customer data or data from suppliers, and thus the key to creating and maintaining competitive advantages.
ENCRYPTED DATA ANALYSIS

Novel cryptographic methods co-developed by us enable the analysis of encrypted data – one of the greatest breakthroughs in modern cryptography.

Private Set Intersection

Our protocols allow you to capture similar database records from different companies without having to share the database.

Joint Evaluations

Our methods open up new ways for joint evaluations without disclosing your data.

Quantum Safe Applications

We develop secure solutions that are safe and resistant to the threat of large-scale quantum computing in the long term.

Novel Cryptographic Solutions

We combine our deep expertise in symmetric cryptology development and cryptanalysis with advanced cryptographic approaches.

Quantum Information

We fuse our expertise in cryptography and privacy-friendly analytics with quantum information.