Execution Environments & Innovative Computing Methods

What types of AI environments exist? How can they be secured against attacks? How can AI execution environments be integrated into security-critical systems? These and similar questions are addressed by our scientists in the Execution Environments department.

When working in cloud/edge-based system environments such as Gaia-X, it is important that overarching principles for the secure and efficient execution of AI (in software and hardware) are developed. These are conceptualised through a largely application-agnostic approach, enabling them to be easily projected onto actual implementations with differing performance requirements. Work is also being carried out on highly specialised environments which enable a high level of reliability in specific application environments - for example, in various aerospace applications and in automated systems. Further research activities will result from the regulatory and legal requirements that AI systems will have to meet in practical applications or in safety-critical areas of application.

In order to permanently embed artificial intelligence in technological maps - without compromising on attack security and robustness - the entire system on which the AI is executed must be considered. To this end, this department also works on securing AI agents, e.g. at the software platform levels for the permanent integration of AI software modules in vehicle technology, as in the Gaia-X 4 AGEDA project.

Quantum computing and other innovative computing approaches are an important aspect of the design of future execution environments. We analyse the possibilities and consequences that new computing methods have for artificial intelligence. Specifically, the questions arise, for example, as to how secure and comprehensible AI can be realised on quantum computers, what opportunities the quantum world offers regarding the development of new methods beyond the improvement of classical algorithms, how new approaches can be embedded in existing classical processes and structures, and how attack scenarios on our IT infrastructures and defense mechanisms will look in the future.