---
id: 762
title: "AI vulnerability deep dive: Prompt injection"
source_domain: bugcrowd.com
source_url: "https://www.bugcrowd.com/blog/ai-vulnerability-deep-dive-prompt-injection/"
published_at: "2024-04-23 12:00"
tags: ["ai-security", "prompt-injection", "vulnerability", "llm", "bugcrowd", "owasp"]
phying_url: "https://phying.com/item/762"
lang: en
---

# AI vulnerability deep dive: Prompt injection

## Summary

Bugcrowd's deep dive on prompt injection: untrusted user input blended into LLM prompts, much like SQL injection, can cause data breaches; mitigations are improving but it may never be fully solved.

## Intelligence Metadata

- **Sources:** bugcrowd.com
- **Published:** 2024-04-23 12:00
- **Original:** https://www.bugcrowd.com/blog/ai-vulnerability-deep-dive-prompt-injection/
- **Phying News:** https://phying.com/item/762