Description
In this work, we address the critical yet understudied question of the security of the most widely deployed pseudorandom number generators (PRNGs) in AI applications. We show that these generators are vulnerable to practical and low-cost attacks. With this in mind, we conduct an extensive survey of randomness usage in current applications to understand the efficiency requirements imposed in practice. Finally, we present a cryptographically secure and well-understood alternative, which has a negligible effect on the overall AI/ML workloads. More generally, we recommend the use of cryptographically strong PRNGs in all contexts where randomness is required, as past experience has repeatedly shown that security requirements may arise unexpectedly even in applications that appear uncritical at first.
Prochains exposés
-
Encryption homomorphe sans bruit à l'aide de groupes
Orateur : Pierre Guillot - Ravel Technologies (dispo Université de Strasbourg, IRMA)
Je vais rappeler les travaux de Nuida et Ostrovski sur l'utilisation des groupes pour l'élaboration de schémas cryptographiques homomorphes. Je vais présenter nos travaux qui fournissent des encodages à la fois plus efficaces et plus généraux, et qui déterminent exactement quels groupes peuvent être utilisés. Puis je vais discuter GRAFHEN, un protocole qui utilise ces idées. Je dirai juste[…]-
Cryptography
-