The field of AI security and risk assessment is rapidly evolving, with a growing focus on addressing the unique challenges posed by adversarial machine learning threats and cognitive cybersecurity vulnerabilities. Researchers are developing innovative frameworks and methodologies to quantify and mitigate these risks, such as comprehensive risk assessment frameworks and vision-first architectures for robust web agents. Noteworthy papers in this area include CIA+TA Risk Assessment for AI Reasoning Vulnerabilities, which introduces a quantitative risk assessment methodology for cognitive security risks, and FRAME, which presents a comprehensive and automated framework for assessing adversarial machine learning risks. Additionally, WebSight: A Vision-First Architecture for Robust Web Agents proposes a novel architecture for interacting with web environments through visual perception, and CORTEX introduces a multi-layered risk scoring framework for assessing AI system vulnerabilities.