About Badcompany

A realistic red-team game to help researchers and enthusiasts find prompt-injection and social engineering flaws in LLM-based agents.

Mission

We create playful but realistic agent simulations so that vulnerabilities can be discovered in a safe environment.

Our Vision

Securing the future of AI by understanding and mitigating real-world attack vectors on autonomous agent systems.

Stay Updated

Get the latest on AI security research and product updates