Research question
Which security risks arise specifically from agent autonomy rather than model behaviour?
Key findings
- Organises risks by autonomy level, memory, tool use and multi-agent interaction - Maps mitigations and open problems
A Survey on Autonomy-Induced Security Risks in Large Model-Based Agents
A comprehensive 2025 survey of security risks created by agent autonomy, memory and tool use.
Contribute
Spot an error, or know a record we should add? Members can propose additions and submit corrections.