OWASP GenAI Security Project Team

LLMs can’t forget what they’ve seen, but you can stop them from saying too much. Here’s how.

From source code to secrets, AI can expose more than you expect, especially when business speed outpaces security readiness.

Attackers aren’t breaching firewalls — they’re slipping instructions into prompts. Here’s why OWASP named prompt injection the top GenAI risk, and what it means for security teams.

You can’t patch prompt injection, but you can outsmart it. OWASP’s latest guidance lays out a layered defense strategy for building safer, more resilient GenAI applications.