Which mechanism prevents public exposure when floating IPs are attached to permissive security groups?

In OpenStack Neutron, security groups provide stateful filtering at the port level. While the default behavior restricts ingress traffic, a common risk involves instances assigned floating IPs that are associated with security groups containing permissive rules (e.g., 0.0.0.0/0 for administrative ports).

Because security groups are applied at the instance port, they do not provide a global subnet-level safeguard to override accidental permissive rules defined by users. This creates a dependency on individual port configurations to prevent unintended public exposure.

Does OpenStack provide a native, stateless subnet-level filtering mechanism that can act as a mandatory guardrail regardless of the port-level security group settings? If so, how does the priority between subnet-level ACLs and port-level security groups resolve when conflicting rules are present?