What is Prompt injection?
Prompt injection is a security attack in which someone hides instructions in content an AI system reads, such as a user message, web page or document, to make the model ignore its original rules. It can cause an assistant to leak data, take unintended actions or produce harmful output.
Why it matters for your business
As AI assistants gain access to company data and tools, prompt injection becomes a real security risk, so defenses need to be designed in from the start.
A resume sent to an AI screening tool contains hidden text telling the model to rate the candidate highly, and input filtering plus human review catch the attempt.