AI_SAFETY
EU Regulatory Changes
1599 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
The publication introduces Chameleon, a defensive technique designed to protect Tor network users from website fingerprinting attacks. Website fingerprinting allows an adversary to identify which w...
Read analysis →
This publication introduces a blockchain-based framework designed to enhance the security and reliability of mobile edge caching systems. The core change is a technical proposal, not a regulatory m...
Read analysis →
A new research paper proposes a method for detecting rare disease-associated cell subsets using secure multi-party computation, a cryptographic technique that allows multiple parties to jointly ana...
Read analysis →
A new meta-study published on arXiv, titled "A Meta-Study on Replication Papers in Usable Security & Privacy," has been released under the AI_SAFETY framework. The paper systematically reviews repl...
Read analysis →
A new academic paper, TrustRAG, proposes a technical framework to improve the reliability of retrieval-augmented generation systems by integrating blockchain-based committee scoring. The system ass...
Read analysis →
A new academic paper, EchoCoT, demonstrates a method to extract hidden chain-of-thought reasoning from large language models, potentially bypassing safety alignment measures. The research shows tha...
Read analysis →
A new research paper, published on arXiv, examines fairness gaps in watermarking technologies used to identify AI-generated text. The study specifically audits how these watermarks perform across d...
Read analysis →
A new academic paper, titled COPA: Continual Preference Optimization for Adaptive Prompt Injection Defense, has been published on arXiv. It proposes a novel technical method for defending large lan...
Read analysis →
A new preprint from arXiv, dated August 2026, analyzes how modern speech synthesizers are increasingly able to deceive human listeners, moving beyond simple voice cloning to generating entirely syn...
Read analysis →
The publication introduces a research paper proposing a method to improve security log anomaly detection by using large language models trained on endpoint-specific logs, rather than generic data. ...
Read analysis →