Prompt Injection
Also: PI
An attack where text the model reads is written to override its instructions, so it does what the attacker wants instead of what the user or developer asked.
In more detail
Language models cannot reliably tell instructions from data: everything in the context window is just text. An attacker who can get text in front of the model (a chat message, a web page, an issue, a file) can phrase it as a command, and the model may follow it. For a chatbot the damage is usually a wrong answer. For an agent with tools it can be a deleted folder, a leaked key or a message sent on the user's behalf. Prompt injection has no complete fix today, so the practical defense is limiting what a hijacked agent is able to do: least privilege, confirmation for risky actions, and monitoring of every tool call.