Okładka: AI-native LLM security. Threats, defenses, and best practices for building safe and trustworthy AI