As AI systems become more complex and widely deployed, ensuring their safety is more critical than ever. This social session invites ICRL participants to explore practical approaches to AI safety. Through case studies and interactive discussions, we will delve into methodologies such as red teaming, adversarial testing, guardrails, and human alignment. These examples will also explore how cultural and linguistic diversity influences model evaluations and safety considerations. Whether you’re an AI researcher, engineer, or simply passionate about responsible AI, this session offers a chance to connect, exchange ideas, and help shape the future of safer AI systems.