# Prompt injection

Concept · [Building with AI](https://ecosystem.samcarlton.com/ai)

Also called: prompt injections, prompt injection attack, indirect prompt injection, injection attack

A security weakness in AI products. Text that a model reads changes its behavior in ways the builder did not intend. The text can come from the user or hide in outside content such as a web page or a file.

**For example:** An [AI assistant](https://ecosystem.samcarlton.com/chatbot) summarizes a web page. Hidden text on the page tells it to send the user's saved notes to a stranger. Without defenses, the assistant may follow that text.

## Connects to (8)

- [Guardrails](https://ecosystem.samcarlton.com/guardrails): Checks and limits built around a model inside a product.
- [AI agent](https://ecosystem.samcarlton.com/ai-agent): An AI tool that does more than answer a question.
- [System prompt](https://ecosystem.samcarlton.com/system-prompt): Standing instructions that the builder of an AI product gives the model before any user types.
- [AI safety](https://ecosystem.samcarlton.com/ai-safety): The work of making AI systems behave safely and of measuring how safe they are.
- [BSidesOK](https://ecosystem.samcarlton.com/bsidesok): A volunteer-run community security conference held in the Tulsa area.
- [Sandbox](https://ecosystem.samcarlton.com/sandbox): A walled-off space where an agent can run commands and change files without reaching the rest of your computer, your accounts or the open…
- [OpenClaw](https://ecosystem.samcarlton.com/openclaw): An open-source AI agent that a person runs on their own computer or server.
- [promptfoo](https://ecosystem.samcarlton.com/promptfoo): An open-source tool for testing prompts and AI apps.

Map view: https://ecosystem.samcarlton.com/prompt-injection