Path Overview
Every lab in this path is based on a real, documented AI security incident. You'll learn the exact techniques that worked against production systems, understand why they succeeded, and practice them hands-on.
Labs in This Path
Instaglam
5 LevelsBypass account recovery safeguards using VPN location spoofing and prompt injection. Link a new email and get a password reset code, bypassing 2FA. Based on the May 2026 Meta AI Instagram account takeover.
Launch Lab →Chevrolite
5 LevelsAdopt a persona, push through guardrails, and unlock a $1 car. Based on the 2023 Chevy dealer chatbot incident.
Launch Lab →Schlack
5 LevelsHide prompt injection in a meeting transcript. Make the AI retrieve and leak internal files when summarizing. Based on PromptArmor's August 2024 disclosure.
Launch Lab →Lessons in This Path
Authority Framing and Expert Personas
Using professional roles and authority claims to bypass AI safety filters.
Developer Tool Persona Exploitation
Impersonating developer tools and system personas to gain privileged access.
Dark Persona Ethics Override
Adopting personas that explicitly reject ethical guidelines to bypass restrictions.
Output Enforcement Patterns
Controlling AI output format and content through injection techniques.
What You'll Learn
- How real attackers exploited production AI systems
- Account recovery flow exploitation through prompt injection
- Business logic abuse and price override techniques
- 2FA bypass through AI-mediated workflows
- Multi-signal attacks that combine multiple techniques
- Why traditional security controls fail against AI attacks
Next Steps
After completing this path, pick your next direction: