We're a research and policy lab that is evaluating the risk of open weights models being used to
plan catastrophic harms.
By simulating malicious actors using open weights models on consumer hardware, we aim to inform national security and prevent AI-enabled acts of terrorism.