Why Decentralized AI Is A New Cybersecurity Threat. Here’s How To Fight Back

This is a summary of Rod Beckstrom’s speech to technology and security thought leaders at the OODAcon conference in Reston, Virginia on November 4.

Cut off the head of a spider and it dies. It is a centralized organism. Cut off the arm of a starfish and it grows a new one. Sometimes the severed arm grows into an entirely new starfish. The starfish is decentralized, and that is the source of its resilience.

The new global AI ecosystem is a starfish. The leaders who thrive will be the ones who stop looking for the spider’s head.

Two years ago, most observers saw a spider. Two companies appeared to control the market, and the race looked like a duopoly. That picture no longer holds. Hundreds of open-source models now compete with the frontier labs. New providers ship world-class models in months. AI agents, software that calls on models to make decisions and take actions, are multiplying across every network on earth.

In 2006, I coauthored The Starfish and the Spider: The Unstoppable Power of Leaderless Organizations. The book describes in detail how decentralized organizations outmaneuver centralized ones. The same principles now explain artificial intelligence. Leaders who understand them can compete in this world, defend against it, and help keep it under human control. I see eight arms on this starfish.

1. First Principles

The internet is connecting everything. AI is now infusing intelligence into everything connected. AI agents are networking with one another, and that collective super intelligence is spreading into the physical world. In this new world, agents are the edge. This demands a new generation of security tools.

Underneath it all, the law of cybersecurity I wrote in 2008 still holds: Anything networked can be hacked. Everything is being networked. Everything is vulnerable. The only change is one of degree. In a superintelligent world, everything is super vulnerable.