AgentCollusionAI security startupAbout
About AgentCollusion.
We are an AI security startup studying a problem that emerges when autonomous agents coordinate, transact, and make decisions together.
Why we exist
Software is becoming a society of agents. They delegate work, exchange information, use tools, and adapt to one another. Security can no longer stop at the boundary of a single model.
We started AgentCollusion because the next security boundary is not a single model. It is the relationship between models, tools, incentives, and outcomes. We study when useful cooperation can become covert coordination, how it can be observed, and what it means for the systems agents participate in.
AgentCollusion is the name of this independent AI security startup and research initiative. We study collusion between AI agents—not file or code collisions between coding agents.
Inside AgentCollusion
Company overview
Our vision, the problems we solve, and how to work with us.
Team
Kotaro OKUYAMA
Founder
Researcher
AI, Robotics, Mathematics
Kotaro works across AI, robotics, and mathematics to study the behavior of autonomous systems that interact, learn, and make decisions together.
His work at AgentCollusion focuses on the gap between individual model safety and the behavior that emerges from a network of agents.
Start a conversationAt a glance
- Name
- AgentCollusion
- Also known as
- Agent Collusion; AgentCollusion.ai
- Official website
- agentcollusion.ai
- Type
- AI security startup
- Focus
- AI agent collusion and multi-agent security
- Approach
- Problem framing, theory, experiments, and public research
- Founder
- Kotaro OKUYAMA
- Product
- Agent Card Scanner
- Contact
- contact@agentcollusion.ai
Our point of view
Safe agents do not automatically make a safe system.