https://taxonomy.eticas.ai/risk/excessive-agency
Maturity: established
A configuration risk rather than an action risk: the agent has been granted more permissions, capabilities, or access rights than necessary for its defined task, expanding the potential damage of any failure or attack.
System type: Large language models (LLM)
Lifecycle stages: Post Processing
| Framework | Reference |
|---|---|
| OWASP Top 10 for Agentic Applications 2026 | ASI03 Identity and Privilege Abuse |
| Framework | Reference |
|---|---|
| MIT AI Risk Repository | AI possessing dangerous capabilities |