Address approval flooding, fatigue, and effects permitted without approval

Investigate defenses against approval flooding and approval fatigue, and determine how to prevent harmful effects that existing policy permits without requesting approval.

Background

The paper reports that ASU's flood controls reduce but do not eliminate attack success against an imperfect approver. The evaluation also shows that approval fatigue depends on the number and selection of prompts, while effects allowed by policy bypass approval entirely. The conclusion explicitly identifies these limitations as remaining open problems rather than claiming that effect binding and trusted rendering resolve them.