When user input contains hidden instructions that override your original prompt, the AI follows the newest directive instead of your intent. This happens more easily than you'd expect and matters more when AI is making decisions on your behalf.
Prompt injection is an attack or accident where unexpected instructions embedded in content, such as a web page or pasted document, override your original instructions and manipulate how the AI behaves.
Understanding this risk helps you use AI safely when processing external content, so you can catch when AI output seems suddenly off-mission or suspiciously redirected.
Peri can explain this concept, give practical examples, help you decide whether it applies to your situation, or recommend a journey if appropriate.
Explore related journeys or tell Peri what you're working through.