[NLU Workbench] When training the NLU Model, it throws the error "Table [ml_labeled_data] is not accessible from scope [x_custom_scope]" and the NLU model cannot be successfully trained and publishedIssue While training an NLU model it throws the error "Table [ml_labeled_data] is not accessible from scope [x_custom_scope]". When providing Test panel feedback in the NLU Workbench, it will create entries in table [ml_labeled_data]. When done on the Development instance, when you train the NLU Model, it will automatically create a cross-scope privilege record with the following message - Read operation on table 'ml_labeled_data' from scope 'x_custom_scope' was granted and added to 'x_custom_scope' cross scope privileges However, if the scoped application with the NLU Model is transferred to the target instance via the App Repo, and then you then provide the Test panel feedback in the NLU Workbench on TEST or PROD, and not in the DEV instance, it will not automatically create the cross-scope privilege record when you train the NLU Model with the feedback, and it will throw the following error message - Table [ml_labeled_data] is not accessible from scope [x_custom_scope]CauseWhen you provide the Test panel feedback in the NLU Workbench on TEST or PROD, and not on the DEV instance after installing the custom app for your Virtual Agent/NLU application on the target instance.ResolutionYou will need to manually create a cross-scope privilege record [sys_scope_privilege] in your custom scope for Virtual Agent/NLU on the target instance to allow read access on the table [ml_labeled_data] in the Global scope, as per the following screenshot - Once created, the NLU Model linked to the Intent Feedback records in table [ml_labeled_data] will be able to successfully complete training without this error and then be published. Ideally, you would provide the feedback using the Test panel feedback in the NLU Workbench on the Development instance, as this will automatically create the cross-scope privilege record that will be part of your custom application, which will then be installed on the target instance and the error will not occur during NLU Model training.