Google PhD Fellowship

Among the 2025 Global Google PhD Fellowships, in the Privacy, Safety, and Security category, is Paul Gerhart. He is a doctoral student at TU Wien’s Privacy Enhancing Technologies research unit, supervised by Dominique Schröder. His work focuses on advanced cryptographic protocols — especially threshold signatures, partially-oblivious pseudorandom functions (PoPRFs), and password-based cryptography — with an emphasis on strong security guarantees and practical efficiency. He holds an M.Sc. in Computer Science from FAU Erlangen–Nürnberg (2022) and a B.Sc. in Mathematics from the University of Bonn (2020).

In 2025 he co-authored four papers presented at flagship conferences — ASIACRYPT 2025, CRYPTO 2025, and PoPETS 2025. At ASIACRYPT, “Password-Hardened Encryption Revisited,” co-authored with Ruben Baecker and Dominique Schröder, uncovers a practical weakness that enables offline password guessing in real systems, then proposes a faster, provably secure redesign with a realistic model including key rotation. Also at ASIACRYPT, “Universally Composable Password-Hardened Encryption,” by an international team of co-authors, identifies a proof error in prior work, introduces the first UC formalization for PHE/TPHE with key rotation, and presents a round-optimal, provably secure, and efficient protocol. At CRYPTO, “A Fully-Adaptive Threshold Partially-Oblivious PRF,” by Ruben Baecker, Paul Gerhart, Daniel Rausch, and Dominique Schröder, delivers a threshold PoPRF with proactive key refresh and composable security, addressing gaps in earlier models and proofs for privacy-preserving applications. At PoPETS, “SoK: Descriptive Statistics Under Local Differential Privacy,” by René Raab, Pascal Berrang, Paul Gerhart, and Dominique Schröder, systematizes LDP methods for means, variances, and frequencies, shows equivalences among common estimators, adds variance estimators, runs empirical comparisons, and offers practical recommendations and caveats for real deployments.

Together, these papers advance both theory and practice: they tighten the foundations of password-hardened encryption, introduce stronger and more usable cryptographic primitives for privacy-preserving systems, and provide evidence-based guidance for deploying local differential privacy. The result is more secure authentication, more robust building blocks for private computation, and clearer best practices for privacy analytics—impacting real-world systems as well as future research.