Faculty
2025: Werner-von-Siemens-Fellow
2025: ERC Starting Grant
2024: GI Junior-Fellow
2024: Busy Beaver Award "Differential Privacy: Mathematical Foundations and Applications in Machine Learning“, Saarland University
Franziska Boenisch is a tenure-track faculty member at the CISPA Helmholtz Center for Information Security. At CISPA, she co-leads the SprintML Lab (Secure, Private, Robust, Interpretable, and Trustworthy Machine Learning), where her research focuses on advancing trustworthy machine learning. Prior to joining CISPA, she was a Postdoctoral Fellow at the Vector Institute for Artificial Intelligence, working under the supervision of Prof. Dr. Nicolas Papernot. She received her PhD from Freie Universität Berlin, where she also served as a research associate at the Fraunhofer Institute for Applied and Integrated Security (AISEC).
National Conference of the American Association for Artificial Intelligence (AAAI) On Stealing Graph Neural Network Models
Association for the Advancement of Artificial Intelligence (AAAI) Demystifying Foreground-Background Memorization in Diffusion Models
International Conference on Learning Representations (ICLR) Curation Leaks: Membership Inference Attacks against Data Curation for Machine Learning
AAAI 2026 Workshop on AI Governance Frequency-Domain Model Fingerprinting for Image Autoregressive Models
ICLR 2026 Workshop: Principled Design for Trustworthy AI
Conference on Neural Information Processing Systems (NeurIPS) Exploring the limits of strong membership inference attacks on large language models
Conference on Neural Information Processing Systems (NeurIPS) Memorization in Graph Neural Networks
National Conference of the American Association for Artificial Intelligence (AAAI) Beautiful Images, Toxic Words: Understanding and Addressing Offensive Text in Generated Images
Conference on Neural Information Processing Systems (NeurIPS) BitMark: Watermarking Bitwise Autoregressive Image Generative Models
Naval Research Logistics Personalized Differential Privacy for Ridge Regression Under Output Perturbation