OWASP Top 10 for LLM Applications (2025)

Top 10 risks for large language model applications: prompt injection, data poisoning, model theft, and more.