Hello, you are using an old browser that's unsafe and no longer supported. Please consider updating your browser to a newer version, or downloading a modern browser.
Training Camp • Cybersecurity Glossary
Prompt injection is an attack that manipulates an AI language model into ignoring its instructions by hiding malicious commands inside user input or external content.
Prompt injection Definition: Prompt injection is an attack that manipulates an AI language model into ignoring its instructions by hiding malicious commands inside user input or external content.
Prompt injection is a technique attackers use to subvert an AI language model by embedding hostile instructions inside the text the model reads, causing it to ignore its original guidelines and follow the attacker's commands instead. It comes in two main forms: direct injection, where the malicious instruction is typed straight into the prompt, and indirect injection, where the instruction is hidden in an external source the model later processes, such as a web page, document, or email. Because the model treats all incoming text as potential instructions, prompt injection can lead to data leakage, unauthorized actions, or the bypassing of safety controls. Defenses include input validation, separating trusted instructions from untrusted content, and limiting what actions a model is permitted to take.
Prompt injection is one of the topics you'll master in the AIGP Boot Camp.
AIGP Boot Camp →