The field of AI safety is moving towards a more comprehensive and systematic approach to evaluating and ensuring the safety of AI systems. Researchers are developing new methods and frameworks for measuring and assessing AI safety, including the use of behavioral techniques, internal techniques, and governance frameworks. These approaches aim to provide a more nuanced understanding of AI capabilities and tendencies, and to translate evaluation results into concrete development decisions. A key challenge in this area is addressing the dual-use dilemma, where the same AI system can be used for both beneficial and harmful purposes. To address this, researchers are exploring the use of access control frameworks and risk-aware security-by-design approaches. Noteworthy papers in this area include: Safety by Measurement: A Systematic Literature Review of AI Safety Evaluation Methods, which proposes a systematic taxonomy for AI safety evaluations. Engineering Risk-Aware, Security-by-Design Frameworks for Assurance of Large-Scale Autonomous AI Models, which presents an enterprise-level approach to AI safety and security. Access Controls Will Solve the Dual-Use Dilemma, which proposes a conceptual access control framework for governing AI capabilities.