The correct answers are A and B . In Zero Trust architecture, policy enforcement is not limited to a plain deny decision. Instead, policy can apply contextual control actions based on the assessed risk of the user, device, session, or application behavior. A conditional block policy is meant to stop or contain malicious or unauthorized activity while also reducing attacker effectiveness.

Quarantine fits this model because it stops access and places the session, user, or device into a controlled state for further review or remediation. That aligns with Zero Trust principles of least privilege, continuous assessment, and adaptive response. Deceive also fits because modern Zero Trust protections can misdirect suspicious or malicious activity toward controlled decoy resources, limiting real exposure while improving detection and response. This is consistent with Zscaler architecture language describing inline prevention, deception, and threat isolation as protective controls.

By contrast, Allow the connection is not a block action, and Firehose is not a standard Zero Trust conditional block control in the architecture concepts you are testing against. Therefore, the two correct answers are Quarantine and Deceive.