Your guide to common prompt defenses
bugcrowd.com | blog | #ai-security | #prompt-injection | #llm | #prompt-defense | #input-sanitization | #adversarial-inputs
Summary
This guide reviews common LLM prompt defenses—input preprocessing, sanitization, blocklists, and allowlists—and explains why non-deterministic model behavior lets attackers bypass them.
- Published
- Collected
--- id: 700 title: "Your guide to common prompt defenses" source_domain: bugcrowd.com source_url: "https://www.bugcrowd.com/blog/your-guide-to-common-prompt-defenses/" published_at: "2024-11-14 16:00" tags: ["ai-security", "prompt-injection", "llm", "prompt-defense", "input-sanitization", "adversarial-inputs"] phying_url: "https://phying.com/item/700" lang: en --- # Your guide to common prompt defenses ## Summary This guide reviews common LLM prompt defenses—input preprocessing, sanitization, blocklists, and allowlists—and explains why non-deterministic model behavior lets attackers bypass them. ## Intelligence Metadata - **Sources:** bugcrowd.com - **Published:** 2024-11-14 16:00 - **Original:** https://www.bugcrowd.com/blog/your-guide-to-common-prompt-defenses/ - **Phying News:** https://phying.com/item/700