Prompt Injection

An attack that attempts to override an AI system’s intended instructions.

Prompt injection occurs when malicious or untrusted text attempts to change how a model follows its original instructions. The attack may be written directly by a user or hidden inside a webpage, email or retrieved document.

Defences include isolating instructions from data, limiting tool permissions, validating actions and treating external content as untrusted. Prompt wording alone cannot provide complete protection.

Also in

Diffusion Models Small Language Models Retrieval-Augmented Generation Prompt Engineering Context Engineering Few-Shot Learning

Apply

Begin the first module

Become AI native, it is the real deal today, and if it is not for you, you have lost nothing but learnt a new skill.

Apply