KI-Agenten & Sicherheit: Franziska Boenisch individualisiert Privatsphäre im Maschinellen Lernen

“Our goal is not to program a huge system, but to use the existing system in the best possible way to address our research questions,” says Franziska Boenisch, Tenure-Track Faculty member at the CISPA Helmholtz Center for Information Security and Werner von Siemens Fellow 2025. In the seventh episode of the Werner von Siemens Ring Foundation's podcast “Ring of the Future,” she discusses her work on trustworthy AI and individualized privacy in machine learning. Join us in this episode for groundbreaking ideas on how our society can harness the opportunities of artificial intelligence without compromising sensitive data—for example, through watermarking or security guarantees. Learn how Franziska Boenisch, who recently received a prestigious ERC grant for her research—a highly sought-after European funding program for key research areas—views the potential developments of AI, such as AI agents that accompany us in everyday life, and how she herself strikes a balance between data protection and trust. Visit our website or follow us on social media if you would like to learn more about the foundation: Website: https://siemens-ring.de/ LinkedIn: stiftung-siemens-ring Bluesky: @siemens-ring.bsky.social YouTube: @werner-von-siemens-ring Franziska Boenisch: https://cispa.de/de/people/c01frbo About Franziska Boenisch's award as a Werner von Siemens Fellow: https://siemens-ring.de/preistraeger/... Would you like to leave feedback? Use the following link to help us further develop the podcast: https://app11.vdivde-it.de/goto/RingD... 0:00 Intro 1:28 Field of work 2:17 Role models 4:02 Questioning AI has more implications than Googling 7:16 Technology companies invest in research 9:21 Data memorization 10:51 Does AI have preferred learning styles? 13:19 Stress tests for AI systems 14:16 ERC grants for security guarantees 15:58 Werner von Siemens Ring Fellowship 23:25 AI systems of the future 27:03 The sheer volume of AI content in science 29:00 Starting point: Watermarking 32:47 The future of AI is philosophical in nature 35:16 How trustworthy can AI become? 38:18 AI responses depend on values ​​in the respective context