Theme of inquiry
What security challenges emerge from conflicting incentives in agent systems?
A question within its area, explored through 3 lines of inquiry below — each a family of specific questions the research asks.
65 specific questions
- How much does interface design shape privacy behavior versus underlying privacy concerns?
- Does giving users more control over AI agents actually improve privacy protection?
- How do false mental models about AI systems affect user privacy behavior?
- Does giving users memory control actually reduce their privacy concerns?
- What share of stated privacy concerns reflect actual platform behavior versus misconceptions?
- Why do people with high AI trust fail to notice privacy leaks?
- How do privacy concerns translate into catching actual information breaches?
64 specific questions
- How do authorization boundaries differ between human and AI agent contexts?
- Can companies control what access levels different AI agents receive across systems?
- How do power users differ in granting autonomy to AI agents?
- How do graduated autonomy levels differ from traditional role-based access control?
- How do identification, authentication, and authorization map onto agentic systems?
- What gaps remain when applying OAuth to agentic delegation?
- What mechanisms enforce authority link constraints when agents run?
67 specific questions
- Can outcome-only safety reporting hide which layer actually contained an attack?
- How does outcome-only reporting hide a filter's role in safety results?
- Can outcome-only safety reports hide dependencies on server-side filtering rather than alignment?
- Does outcome-only reporting hide which layer actually blocked an attack?
- Can outcome-only reporting hide whether agent safety comes from server-side filters or the model?
- Does provider-side filtering hide true safety from outcome-only attack reports?
- What makes a security boundary evaluation cautious rather than a certification?