Currently free during beta - premium features coming soon. Subscribe now to lock in early access.
AI_SAFETY

EU Regulatory Changes

1476 changes tracked across 24 compliance frameworks including DORA, NIS2, GDPR, EU AI Act, Cyber Resilience Act, and more.

All DORA NIS2 GDPR CSRD MaRisk ISO27001 EU_AI_ACT CRA DSA DMA eIDAS2 SOC2 PCI_DSS HIPAA ISO42001 AMLD6 PSD3 DATA_ACT GPSR CER EUDR CVE BREACH AI_SAFETY
arXiv: SkillJack: Persistent Skill Backdoors in Self-Evolving Agents
arXiv: SkillSentry: Adaptive Honey Worlds for Dynamic Safety Testing of Agent Skills
arXiv: Breaking ACDGV MinRank Gabidulin encryption schemes over matrix codes
arXiv: Provably Learning Multi-Head Attention with Queries
arXiv: Noise-Aware Shrinkage for Differentially Private Zeroth-Order Fine-Tuning of Large Language Models
arXiv: Right Divisibility in Erasing Semi-Thue Systems: A Minimal View of Intruder Deduction
arXiv: Attacking and Defending Multi-Agent Collaborative Filtering Systems Through Connectivity
arXiv: FedGSA: Geometry-Consistent Subspace Aggregation for Differentially Private Federated LoRA
arXiv: ShielDroid: A Hybrid Approach Integrating Machine and Deep Learning for Android Malware Detection
arXiv: MalTotal: Cost-Effective and Language-Agnostic Malicious Code Poisoning Detection for Millions of Repositories
arXiv: Attribute-based Undetectable Watermarking for Generative AI Models
arXiv: Test-time reasoning effort and unauthorized tool use in language-model agents: a prespecified equivalence study
arXiv: Solving the Shortest Vector Problem in time $2^{0.6039n}$ Time via Mid-point Hessian
A new academic paper, published on arXiv on August 3, 2026, claims a major algorithmic breakthrough in lattice-based cryptography, specifically solving the Shortest Vector Problem (SVP) in time 2^(...
Read analysis →
arXiv: Agentic Incident Response through Digital Twin-Enhanced Multiscale Planning
A new research paper, published on arXiv, proposes a technical framework for using AI agents to manage incident response, specifically by creating digital twins of critical infrastructure to run mu...
Read analysis →
arXiv: Antares: Foundation Models for Agentic Vulnerability Localization
A new research paper, Antares: Foundation Models for Agentic Vulnerability Localization, has been published on arXiv, presenting a novel AI system designed to automatically identify and locate soft...
Read analysis →
arXiv: Self-Supervised Representations for Binary Program Clustering: From Empirical Study to Retrieval-Augmented Lea...
This publication introduces a new machine learning technique that groups similar compiled software programs, known as binaries, by analyzing their underlying structure without needing human-labeled...
Read analysis →
arXiv: Lost in Permissions: Exploring the Microsoft 365 App Ecosystem
A new academic study, published on arXiv in August 2026, examines the security and privacy risks within the Microsoft 365 app ecosystem, specifically focusing on how third-party applications reques...
Read analysis →
arXiv: TrainShield: Targeted Awareness for Cybersecurity Training
TrainShield is a newly published research paper proposing a targeted cybersecurity training framework that uses AI to personalize awareness programs based on an individual’s actual risk exposure an...
Read analysis →
arXiv: A Multi-Objective AutoML-based Efficient Intrusion Detection System for EV Charging Networks
A new research paper proposes a machine learning framework for detecting cyber intrusions in electric vehicle charging networks. The study introduces an automated, multi-objective system that optim...
Read analysis →
arXiv: TurboRetry: Mitigating Large-Scale QUIC Handshake Floods with Off-the-Shelf DPU Offloading
This publication introduces TurboRetry, a technical framework designed to mitigate large-scale QUIC handshake floods by leveraging off-the-shelf Data Processing Units (DPUs). It is not a regulatory...
Read analysis →