EU Regulatory Changes

The publication, titled "Formalization of security" under the AI_SAFETY framework, introduces a rigorous mathematical and logical structure for defining and verifying security properties in AI syst...

A new academic paper proposes using homomorphic encryption to implement logic locking in system-on-chip designs, a technique that could allow hardware to be securely activated or deactivated post-m...

A new research paper, published on arXiv on July 30, 2026, introduces a method for improving cybersecurity threat detection using large language models with advanced reasoning capabilities. The stu...

This publication is a research paper from arXiv, not a binding regulatory change, but it offers critical guidance for compliance teams navigating the emerging field of generative AI. The paper exam...

The publication introduces a new technical framework, Encryption-Compatible Clustered Federated Learning via Distributed Expectation-Maximization over Metadata. This is a research paper, not a regu...

This publication is a technical research paper, not a regulatory change, but it has significant compliance implications for hardware security. The paper presents a new framework for understanding a...

A new academic paper, published on arXiv, provides a comprehensive lifecycle analysis of security threats and defenses for world-model-based embodied AI systems. This is not a new regulation, but a...

A new research paper proposes a technical framework for applying secure aggregation to federated learning models trained on clinical electroencephalogram (EEG) data. Federated learning allows multi...

A new academic paper, published on arXiv, demonstrates that simple image transformations, such as slight rotations, color shifts, or compression, can reliably bypass modern AI-based content moderat...

A new research paper, published on arXiv, explores the use of technology-enhanced tabletop exercises for cybersecurity education, with direct implications for AI safety compliance. The study presen...