EU Regulatory Changes

A new preprint from arXiv, titled "Do Modern Post-Hoc Watermarking Methods Beat Broken-Arrows?" published on May 26, 2026, evaluates the robustness of current AI-generated content watermarking tech...

This paper, published on arXiv, introduces BAIT, a new technical framework for improving the safety of large language models (AI systems). BAIT stands for Boundary-Guided Disclosure Escalation via ...

This paper, published on arXiv, presents research on a hidden cost associated with a specific technique used to make large language models (LLMs) forget or "unlearn" problematic data, such as copyr...

A new research paper, "Lessons from Penetration Tests on Large-Scale Agent Systems," has been published on arXiv, detailing systematic security vulnerabilities found in autonomous AI agent systems....

This paper, published on arXiv, presents a new evaluation framework for detecting prompt injection attacks against large language models. The key finding is that no single detection method works un...

A new academic paper published on arXiv, titled "The Fault in Our Drafts: Vulnerabilities in RPKI Specification and Software," has identified critical security flaws in the Resource Public Key Infr...