Theme 01
Agent assurance and multi-agent security
Research on how autonomous agents can be audited, constrained, and kept resilient when they interact with other agents and external systems.
Questions under study
- How should security requirements be stated for agents that plan and act?
- Which forms of verification remain useful when agent behaviour changes over time?
- How do cooperation and adversarial behaviour coexist in multi-agent systems?
- What evidence should an audit preserve after an agent causes harm?