Prompt injection
Text that makes an AI model do something its owner did not intend.
A Prompt Injection Vulnerability occurs when user prompts alter the LLM's behavior or output in unintended ways.
OWASP adds that the text need not be readable by a person: injected inputs "do not need to be human-visible/readable, as long as the content is parsed by the model". An agent that reads outside content or has tools is more exposed, because injected text can reach its tool calls.
What guardrails catch and miss: Can you stop an AI agent being tricked?