Excessive Agency
An AI Agent security risk in which the Agent holds permissions beyond what is necessary, such as shell execution without an allowlist, file access without path validation, or operations performed without human approval.
Timeline (last 90 days)
The OWASP LLM risk list includes the Excessive Agency risk, namely that a model is given too much autonomous decision-making and execution capability without human oversight.
Over-agency refers to an agent being granted permissions, capabilities, or autonomy beyond what is needed to complete a task, leading to unintended or harmful side effects.
All Facts (2)
The OWASP LLM risk list includes the Excessive Agency risk, namely that a model is given too much autonomous decision-making and execution capability without human oversight.
50%UnverifiedOver-agency refers to an agent being granted permissions, capabilities, or autonomy beyond what is needed to complete a task, leading to unintended or harmful side effects.
50%