EU Regulatory Changes

This paper, published on arXiv, introduces Hephaestus, a proposed AI system designed to autonomously conduct cybersecurity research, including vulnerability discovery and exploit generation. While ...

This publication, released on 29 June 2026, presents a novel adversarial attack method called the Explainability-Aware Frustum Attack, which targets LiDAR-based 3D object detectors commonly used in...

A new preprint published on arXiv proposes a framework called GTI-mSEMP, which models how malware could be deliberately stimulated to spread more effectively by incorporating attacker and defender ...

This paper, ToolPrivacyBench, introduces a new benchmarking framework designed to evaluate how well large language model agents protect user privacy when using external tools. It specifically tests...

This paper, published on arXiv, presents a novel measurement study of non-interactive SSH attacks against honeypots, which are decoy systems used to detect cyber threats. The research reveals that ...