Tagged: Application Security

Indirect prompt injection: the security risk hiding in your AI assistant’s documents

When an LLM reads untrusted content, that content can attempt to give it instructions. Why retrieval and tools change the threat model.

1 min read