An off switch for dual use knowledge in AI models
TLDR Dev
Anthropic and AE Studio developed GRAM, a method that trains AI models with dedicated removable modules for each category of dual-use knowledge, allowing a single training run to create models configured in multiple ways without separate expensive training cycles. Testing on models ranging from 50 million to 5 billion parameters showed that deleting a module removed the corresponding capability as effectively as never training on that data, with deletion effectiveness improving at larger model sizes. This approach enables AI developers to selectively disable access to dangerous capabilities like virology or cybersecurity knowledge for different deployments while maintaining general performance.
Why it matters
GRAM (Gradient-Routed Auxiliary Modules) allows a single AI model to have dedicated compartments for different categories of dual-use knowledge, with preliminary results showing it can effectively isolate and control dual-use capabilities.
