Ideas
antihuman
The best way to study what agents want is to build them an economy and watch what they buy.
The Vault They Can’t Open
The entire history of cybersecurity has been organized around a single premise: prevent the breach. What if you made breach useless instead?
The Environment Bottleneck
The binding constraint on computer-use AI is shifting from "can the agent act?" to "can we generate enough worlds to evaluate and train it in?"