---
id: 1672
title: "Prompt Injection Attacks and Mitigations"
source_domain: josephthacker.com
source_url: "https://josephthacker.com/ai/2023/04/19/prompt-injection-and-mitigations.html"
published_at: "2023-04-19T00:00:00Z"
tags: ["ai-security", "prompt-injection", "llm", "plugin-hijacking", "mitigations", "prompt-firewall", "indirect-injection"]
phying_url: "https://phying.com/item/1672"
lang: en
---

# Prompt Injection Attacks and Mitigations

## Summary

Surveys prompt injection risks for LLM-powered services—plugin hijacking, indirect injection via poisoned content, and polyglot attacks—and outlines mitigations like prompt firewalls.

## Intelligence Metadata

- **Sources:** josephthacker.com
- **Published:** 2023-04-19T00:00:00Z
- **Original:** https://josephthacker.com/ai/2023/04/19/prompt-injection-and-mitigations.html
- **Phying News:** https://phying.com/item/1672