Stopping AI swarms from turning malicious.

Dowding Labs studies how teams of AI agents fail, and builds the monitors that keep them safe.

Chain Home radar, 1940

In 1940, Hugh Dowding built a system that saw trouble coming. We’re building one for AI agents.

AI agents now move money, change code and talk to each other with no one watching. Our mission is to make them safe enough to trust.

Named after Hugh Dowding.

By 1940, Air Chief Marshal Hugh Dowding had joined Britain’s radar stations, observers and fighter squadrons into one system. We’re building the same thing for teams of AI agents.

Talk to us.

Running AI agents, or studying them? We’d love to hear from you.