Thinking Machines Lab proposes a staged safety path for open AI models

Original: A Safe Path To Open Weights

Source: thinkingmachines.ai ↗