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

EU Regulatory Changes

1550 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
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: Is External Database Protection Static in Retrieval-Augmented Generation? Rethinking Privacy Preservation unde...
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: FlowGuard: From Signals to Evidence for MCP Security Detection
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: Lattice-based extended withdrawability
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: MemPoison: Uncovering Persistent Memory Threats and Structural Blind Spots in LLM Agents
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: Routing Ceilings Are Domain-Independent: Structural Prior Injection in Code Security Vulnerability Detection
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: Bad Memory: Evaluating Prompt Injection Risks from Memory in Agentic Systems
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: Qubes OS Security in the Public Record
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: Fully Automated End-to-End Adversary Emulation from MITRE ATT\&CK Based Cyber Threat Intelligence Using LLMs
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: Context Contamination in LLM Analysis of Network Security Logs: Poison with Passive Prompt Injection and Mitig...
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: Disclosure Divergence: Measuring Privacy Policy and Data Safety Misalignment at Scale
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: A Measurement Study of AI-Environment Realism Gaps in Malware-Analysis Sandboxes
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: Better Privacy Guarantees for Larger Groups
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: Exploring Delay-based PUFs for Energy-Efficient Low-Overhead Security of Wearable Devices
AI_SAFETY 17 Jul 2026 arxiv_cscr
arXiv: Value Leakage: An LLM's Answers Are Silently Shaped by Its Own Values
AI_SAFETY 16 Jul 2026 arxiv_cscr
arXiv: How Agents Ask for Permission: User Permissions for AI Agents, from Interfaces to Enforcement
This paper, published on arXiv in July 2026, proposes a new technical framework for how AI agents should request and manage user permissions. It moves beyond simple app-style consent popups to a mo...
Read analysis →
AI_SAFETY 16 Jul 2026 arxiv_cscr
arXiv: WarpGuard: Towards Control-Flow Attestation for Heterogeneous CPU-GPU Execution
This publication introduces WarpGuard, a proposed technical framework for control-flow attestation in heterogeneous computing environments where CPUs and GPUs execute code together. Control-flow at...
Read analysis →
AI_SAFETY 16 Jul 2026 arxiv_cscr
arXiv: Protective Capacity Hallucination: When Large Language Models Claim Nonexistent Capabilities
This paper, published on arXiv on July 15, 2026, introduces a new class of AI failure mode termed "protective capacity hallucination." Unlike standard hallucinations where a model invents facts, th...
Read analysis →
AI_SAFETY 16 Jul 2026 arxiv_cscr
arXiv: UTS at ELOQUENT 2026 Voight-Kampff: structural shifts in AI writing bypass state-of-the-art detectors
A new preprint from arXiv, published on July 15, 2026, presents findings from the UTS at ELOQUENT 2026 Voight-Kampff study, demonstrating that recent structural shifts in AI-generated writing can n...
Read analysis →
AI_SAFETY 16 Jul 2026 arxiv_cscr
arXiv: Multivariate Cryptography-Based Anonymous Certificate Scheme
This publication from arXiv presents a new technical proposal for a multivariate cryptography-based anonymous certificate scheme. While not a regulatory change itself, it signals a potential shift ...
Read analysis →
AI_SAFETY 16 Jul 2026 arxiv_cscr
arXiv: When T2I Synthetic Data Backfires: Amplified Privacy Risks in Real-Synthetic Mix Training
This paper, published on arXiv, presents new research demonstrating that training AI image generation models on a mix of real and synthetic data can actually increase, rather than reduce, privacy r...
Read analysis →