# Introduction to Cybral Guard Connectors The Cybral Guard Connectors serve as indispensable tools in easily integrating the Cybral Guard system standards with users' files, so ensuring a strong data classification and security measures. Designed to empower users, Cybral Guard connectors provide the capability to apply classification groups and labels to their files, therefore enhancing data organization, accessibility, and protection. As users navigate their digital environments, these connectors enable them to align with organizational rules and standards for data protection. By understanding the features offered by Cybral Guard connectors, users play a main role in safeguarding sensitive information, maintaining compliance with regulatory requirements, and keeping data security protocols within their organizations. Whether classifying documents or labeling files, users can depend on Cybral Guard connectors to update their workflows and support data protection efforts across all levels of their digital assets. Through the intuitive interface and user-friendly functionalities provided by Cybral Guard, users are empowered to proactively manage and safeguard their data, ensuring its integrity, confidentiality, and availability in agreement with organizational policies and industry best practices. # Users Users' experiences with Cybral Guard will be determined by the three classification modes configured by the administrator on the Cybral Guard portal. These modes include AI classification, manual classification, and recommended classification. Each mode offers distinct functionalities and user interactions tailored to meet the organization's data protection needs. #### AI Classification Mode In the AI classification mode, labels are automatically assigned to files by AI algorithms, restructuring the classification process for users. However, it's important to note that in this mode, while the manual classification group labels will be visible on the top of the file, the labels associated with the AI classification group will not appear. Instead, the AI manages the classification process separately.

If the policy consists only of AI classification groups, users won't see any labels at the top of the file. This is because the AI is responsible for automatically assigning labels based on the content of the files. Thus, users can depend on the AI to classify their files without needing to manually select labels.

#### Cybral Guard AI Classification message When saving the file, a message will appear for the user informing him that the document is undergoing the classification process. After a few seconds, the classification will be completed, and the user will see the labels applied by the AI on the file.

To check the AI classification on a file, simply right-click on the file, select "Properties," and then navigate to the "Cybral" tab. There, you'll find the label assigned by the AI.

#### Manual Classification Mode In Manual Classification Mode, users have the independence to personally assign labels to files, providing the option to file classification. Unlike AI Classification Mode, where labels are automatically assigned by algorithms, users take charge of categorizing their files according to predefined criteria. In this mode, users can visibly see the labels associated with manual classification groups displayed atop the file. In the example below, users will manually assign labels to the three classification groups displayed at the top of the file. This process is part of the Manual Classification Mode, where users take responsibility for categorizing files according to predefined criteria.

Word Document with Mandatory Manual Mode

Before proceeding with labeling the file, users must first save the document. This ensures that any changes made to the file are properly recorded and applied. Once the file is saved, users can then proceed with specifying the appropriate label for each classification group according to the organization's policies.

#### Cybral Guard Manual Classification messages * In the provided example, the user has specified the label for each classification group, ensuring that the appropriate classification is applied to the file. This proactive approach aligns with organizational policies and helps maintain consistency and accuracy in data classification across the system.

* When attempting to add a label multiple times, the system will inform the user that the label has already been applied.

* If a user forgets to assign a label to a file, a message warning them to specify the mandatory classification group will appear. This serves as a gentle reminder to ensure that all files are accurately categorized according to organizational policies and standards.

Mandatory Classification Groups Labeling

* In this example, users need to add labels for ISO and NIST classifications because they're mandatory. A message will remind users of this. However, for the non-mandatory group set by the administrator, labeling isn't required. So, users can choose not to label files for this group if they prefer.

* With an empty files, users are unable to label the file. The document must contain content before users can proceed with assigning labels. This requirement ensures that files are properly classified based on their actual content, enhancing the accuracy and effectiveness of the classification process.

#### Classification correction messages * When a user selects a label with lower severity for classified information that requires a more restricted classification, a classification correction message will appear.

* The system notifies the user of the reason for rejecting the chosen label, stressing that the file contains sensitive data. It then lists the invalid labels, regulations, and sensitive data that cannot be applied, ensuring users understand why their chosen classification is not allowed and guiding them towards selecting the appropriate classification for the file.

* Users have the capability to expand each item within the sensitive data section to access detailed information about that specific item. In the example provided, the user expanded the section of the invalid labels that cannot be applied to the file. This includes the low impact label, which is indicated as unsuitable for classification due to its inability to address the sensitivity level of the file.

* When expanding the regulations section, users will be presented with a list detailing the specific regulations applicable to the file.

* The sensitive data found within the file will be listed for the user to review, providing details about the specific information that requires the application of a particular label. This detailed overview enables users to understand the factors driving the classification requirements for the file, ensuring accurate and appropriate labeling based on the presence of sensitive data.

Users can observe a different format of the classification groups at the top of the file, either in a dropdown menu or as a ribbon. In this example, users will interact with the classification groups in the form of a dropdown menu, allowing them to select the appropriate labels for each group manually.

#### #### Recommended Classification Mode In the recommended classification mode, both manual and AI modes are combined, making it mandatory for users. Since AI intervention is inherent in this mode, the mandatory aspect cannot be optional. Labels will be visible at the top of the file, similar to the manual mode. In this example, the user has designated the label "Normal" for this file.

#### #### Cybral Guard Recommended Classification message In this example, our user forgot to label the file before closing the file, so the file was automatically labeled by the AI. This labeling is based on the files that were uploaded to the system for AI training. In this case, the AI is responsible for labeling the file.

Users have the option to hover the mouse over the labels at the top of the file to access detailed information about each label. This feature provides users with additional context and understanding of the classification applied to their files.

Cybral Guard operates within Outlook, offering users the same classification capabilities as other applications. Users can view classification groups at the top of their emails according to the format defined by the administrator. This could involve manual classification by the user, recommended classification combining manual and AI modes, or entirely AI-driven classification. In the example below, users will see the classifications applied by AI.

In Outlook, if a connection issue arises, Cybral Guard will generate a warning message informing the user to review both the message recipient and the message content. This proactive approach helps users identify and address any communication issues effectively.

{% hint style="success" %} For manual and recommended modes, If a user attempts to assign an incorrect label to a file that falls under the restrictions defined in the AI data classification, the system will prevent them from doing so. Instead, the AI will automatically select a more appropriate label based on the content and context of the file, ensuring compliance with the specified regulations and standards. {% endhint %} {% hint style="info" %} Clicking on the help icon next to the labels at the top of the page

will conveniently direct users to the relevant section of the manual, providing them with instant assistance and guidance on how to interact with the labels and classification system. {% endhint %} # Server Scanner Overview The Server Scanner is a simple service designed to operate on Windows servers, ensuring the security and proper classification of the files located on the server. This scanner is not intended for desktop use but is specialized for server environments. Here’s an organized breakdown of how the process works. ### Installation and Initial Setup When the Server Scanner is installed on a Windows server, files on the server will be ready to begin their journey with the new operation. This setup is critical for the scanner to perform its tasks effectively. #### **File Scanning Process** 1. **Initial Data Discovery**: Upon installation, the scanner discovers the files and starts to scan each file on the server. It precisely analyzes every file, one by one. 2. **Classification Request**: After scanning a file, the scanner sends it to the AI system. The AI system is responsible for classifying the file based on pre-defined classification groups. 3. **AI Classification**: The AI evaluates the file and assigns a classification, identifying and categorizing sensitive data within the file, such as Business Information (BII), Business Health Information (BHI), and regulated information. #### **Continuous Monitoring and Rescanning** To ensure the accuracy and up-to-date classification of files: 1. **24-Hour Interval**: The scanner performs a rescan every 24 hours. This routine check ensures that any updates or changes to the files are detected and reclassified if necessary. 2. **Reclassification**: If any modifications are found, the files are reclassified to reflect the latest information. #### **Classification Groups** 1. **AI or Recommended**: The classification groups used by the scanner include AI or recommended categories. The scanner operates with the AI portion of these classifications, disregarding manual inputs. 2. **No User Interaction**: This service runs autonomously, requiring no user intervention. The AI training process is integral to achieving accurate classification, ensuring the system continuously learns and improves. #### **Service Includes Three-Stage Process** The Server Scanner operates through a three-stage process involving the portal, AI, and the scanner itself: 1. **Portal Configuration**: Users configure classification groups and settings via the portal. 2. **Scanner Operation**: The scanner scans all files on the server, referring to the portal’s configuration to determine the necessary classification groups. 3. **AI Classification**: The scanner communicates with the AI, requesting classification for each file. The AI uses trained models to classify the files and returns specific labels from each classification group. #### **Decision and Validation Cycle** 1. **AI Recommendations**: The AI recommends suitable labels for each file based on its classification groups. 2. **Revalidation**: The scanner revalidates the AI’s decisions against the portal’s configuration. This configuration, set by the admin, includes specific restrictions and labels assigned to particular types of information. This ensures that the classifications align with the predefined rules and policies established by the organization. 3. **Final Decision**: Based on the revalidation, the scanner finalizes the classification of each file. This structured approach ensures a detailed and reliable classification of files on Windows servers, maintaining data security # Known Limitations ### Ignore the Old Thread of an Email Feature CYBRAL GUARD introduces a new feature specifically designed for handling email threads within Outlook, known as "Ignore Old Thread of an Email." It aims to improve efficiency by excluding older email threads from being sent to the AI for classification, which can otherwise be time-consuming. This feature functions normally across most systems. However, users running certain versions of Windows, such as build 21H2, may need to execute Outlook with administrative privileges to ensure the feature operates correctly. This requirement is due to the specific coding techniques used to track old threads, which may necessitate elevated permissions in these environments. By acknowledging and addressing these limitations, users can better use the "Ignore Old Thread of an Email" feature to enhance the efficiency of email classification processes without unnecessary delays.