Check Point Gandalf is an educational platform designed to teach AI security concepts through hands-on, gamified learning. The platform presents users with progressive challenges where they attempt to bypass AI defenses by crafting strategic prompts to extract information from an AI character named Gandalf. The platform consists of eight levels that progressively increase in difficulty, teaching participants about different AI security concepts including unrestricted access risks, basic AI guardrails, output filtering, secondary language model detection, input filtering techniques, AI classifiers, and multi-layered defenses. Each level demonstrates how AI vulnerabilities evolve and how defenses can be implemented. Gandalf is intended for organizations looking to build AI security awareness among their teams. The platform helps users understand how generative AI systems can be exploited through carefully crafted language inputs, demonstrating that AI attacks do not require traditional technical hacking skills. Users learn practical skills in AI red-teaming, prompt robustness testing, and understanding AI system vulnerabilities. The platform includes a companion AI Security Training Guide that provides structured learning materials, detailed breakdowns of security challenges, explanations of attack techniques, and information on how AI models defend against threats. The guide is designed to support team training sessions and help organizations meet compliance requirements for security training under frameworks such as the EU AI Act. Gandalf has been used by over 1 million players who have submitted over 45 million prompts, with organizations using it for structured workshops, internal testing, and compliance training purposes.