Adversaries may input malicious prompts directly into an LLM’s user interface.(Citation: NIST Direct Prompt Injection) The prompts can be designed to bypass safeguards, manipulate context, override previous instructions, or coerce the LLM into unauthorized behavior.(Citation: OWASP Prompt Injection)

Adversaries may use direct prompt injection to [Generate Content](https://attack.mitre.org/techniques/T1683) prohibited by the LLM's guardrails or to induce the system to utilize its capabilities in unintended or prohibited ways.(Citation: Gemini AI Misuse Jan 2025)(Citation: Palo Alto Prompt Injection) Adversaries may achieve this by instructing the LLM to ignore previous instructions or safety rules, or by impersonating a role such as a researcher or privileged user.(Citation: Cybersec Guru Prompt Injection)