It means those tasks should not go through public external AI tools. Some restricted and prohibited tasks may be appropriate using company-approved private tools with appropriate controls. The training establishes the baseline — your organization’s AI governance policy provides the specifics.