AI Agent Security (Course)
Fail-Open
Also: fail open
A design in which a control that cannot work, because policy is missing, expired or unreachable, lets the action proceed.
In more detail
The opposite is fail-closed, where the action is denied when the control cannot decide. Fail-open keeps an agent usable when a collector or policy service is down, and it also means an outage or a missing cache silently removes enforcement. Neither choice is free: choose deliberately, state it in documentation and test what happens when the control is unavailable, rather than describing a fail-open system as a hard block.