EU Regulatory Changes

This document, published on arXiv, introduces the Maestro Order, a proposed technical framework for orchestrating the safe deployment of AI models. It is not a regulation but a model-agnostic harne...

This publication, a research paper from June 2026, analyzes the performance impact of confidential computing on NVIDIA's Blackwell GPUs when serving large language models (LLMs). It introduces a co...

This publication introduces BipBipCache, a novel hardware-level encryption technique designed to secure data within a computer’s cache memory while maintaining very low latency. The paper proposes ...

This publication, titled AutoPRAC, presents a new automated method for discovering attack patterns that can bypass PRAC-based Rowhammer defenses in computer memory hardware. Rowhammer is a vulnerab...

This publication, titled "Are Safety Guarantees in Neural Networks Safe? How to Compute Trustworthy Robustness Certifications," presents a critical analysis of existing methods used to certify the ...

A new research paper, HelpBench, published on arXiv, introduces a benchmark designed to evaluate how well large language models provide advice on privacy, safety, and security. This is not a regula...

This paper, published on arXiv, presents a new framework called Burnyard for analyzing malware using artificial intelligence. It is not a regulatory change from a governing body, but a technical pu...

This publication introduces PowerFuzz, a novel black-box firmware fuzzing technique that uses power consumption measurements to detect vulnerabilities in embedded devices without requiring source c...

This publication, FirmCure, presents a novel technical framework for the autonomous and adaptive rehosting of Linux-based firmware, enabling large-scale dynamic analysis of embedded systems. While ...