Prompt injection is when an attacker inserts hidden instructions into text fed to an AI system—often disguised in data the AI is supposed to process. If your personal information passes through an AI system, a prompt injection attack could trick it into leaking your data, ignoring security rules, or revealing what it shouldn't. The attack works because the AI struggles to distinguish between legitimate instructions and malicious ones hidden in the data.
Prompt injection is an attack where malicious instructions are embedded in content that an AI system reads — such as a webpage, email, or document — causing the AI to perform unintended actions or expose sensitive information it has access to. As AI assistants are granted more access to personal inboxes, calendars, and files, the risk of prompt injection leaking private data grows substantially.
Recognizing how prompt injection works allows users to set safer boundaries when granting AI tools access to personal data, choose platforms with injection-resistant architectures, and understand why reviewing AI permissions regularly is now a core digital privacy habit rather than an optional precaution.
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.