← All resources

Week 3 — Technical Safety Approaches

This week is about the tools researchers are using to make AI systems easier to understand and control. We’ll compare approaches such as interpretability, evaluations and red teaming, scalable oversight, and control—and ask what each one can realistically catch.