LLM Safety
9 practice items · needed by 2 roles
Easy 3Medium 1Hard 5
Practice questions
- What prompt injection iseasy
- Instructions hidden in a retrieved pagehard
- Jailbreak or prompt injectioneasy
- Customer data in promptseasy
- Checking replies before users see themmedium
- Injection through a web pagehard
- Leaking data through an image linkhard
- Hundreds of fake dialogueshard
- API key in the system prompthard