EU Regulatory Changes

This paper, published on arXiv on June 10, 2026, introduces a novel cybersecurity vulnerability called the "Amnesia" attack, which targets continual learning systems. Continual learning is a machin...

This publication introduces a novel technical framework for evaluating the defensibility of AI systems against adversarial manipulation, moving beyond traditional runtime enforcement methods. The p...

This paper, published on arXiv, proposes a new framework for defining a "reasonable standard of care" for cybersecurity, moving beyond simple compliance checklists. It argues that current regulator...

This paper, published on arXiv, introduces a new algorithm for differentially private hierarchical heavy hitters, a technique used to identify the most frequent items in a dataset while preserving ...

This publication from arXiv introduces a new design framework for cryptographic APIs that prioritizes intent-based programming, enabling systems to automatically adapt cryptographic algorithms and ...

A new academic framework has been published on arXiv titled "An Assessment Framework for Application-Level Cryptographic Agility," which proposes a structured methodology for evaluating how easily ...

This paper, published on arXiv, introduces a new benchmarking framework called "Who Pays the Price?" designed to evaluate how real-world web agents—AI systems that interact with websites and online...

This publication introduces a novel cryptographic method called Split Tallies, which enables the auditing of dynamic ordered sets, such as transaction logs or supply chain records, using only const...

This publication, a longitudinal study from arXiv, analyzes how Android malware has increasingly used covert communication channels—such as steganography, encrypted payloads in network traffic, and...

This paper, published on arXiv on June 11, 2026, presents research demonstrating that large language model-powered AI systems can now autonomously develop and execute penetration testing capabiliti...