Researchers propose a safety framework for AI agents using Scala 3 with capture checking to track capabilities and prevent unsafe behaviors like information leakage and unintended side effects. The type system enforces fine-grained control over agent actions while maintaining task performance.