Jens Müller

Peter Lorenz, Mario Fernandez, Jens Müller et al.

Detecting out-of-distribution (OOD) inputs is critical for safely deploying deep learning models in real-world scenarios. In recent years, many OOD detectors have been developed, and even the benchmarking has been standardized, i.e. OpenOOD. The number of post-hoc detectors is growing fast. They are showing an option to protect a pre-trained classifier against natural distribution shifts and claim to be ready for real-world scenarios. However, its effectiveness in dealing with adversarial examples (AdEx) has been neglected in most studies. In cases where an OOD detector includes AdEx in its experiments, the lack of uniform parameters for AdEx makes it difficult to accurately evaluate the performance of the OOD detector. This paper investigates the adversarial robustness of 16 post-hoc detectors against various evasion attacks. It also discusses a roadmap for adversarial defense in OOD detectors that would help adversarial robustness. We believe that level 1 (AdEx on a unified dataset) should be added to any OOD detector to see the limitations. The last level in the roadmap (defense against adaptive attacks) we added for integrity from an adversarial machine learning (AML) point of view, which we do not believe is the ultimate goal for OOD detectors.

Jens Müller, Marcus Brinkmann, Damian Poddebniak et al.

We show practical attacks against OpenPGP and S/MIME encryption and digital signatures in the context of email. Instead of targeting the underlying cryptographic primitives, our attacks abuse legitimate features of the MIME standard and HTML, as supported by email clients, to deceive the user regarding the actual message content. We demonstrate how the attacker can unknowingly abuse the user as a decryption oracle by replying to an unsuspicious looking email. Using this technique, the plaintext of hundreds of encrypted emails can be leaked at once. Furthermore, we show how users could be tricked into signing arbitrary text by replying to emails containing CSS conditional rules. An evaluation shows that 17 out of 19 OpenPGP-capable email clients, as well as 21 out of 22 clients supporting S/MIME, are vulnerable to at least one attack. We provide different countermeasures and discuss their advantages and disadvantages.