Blog Post

Security, Compliance, and Identity Blog
5 MIN READ

New Purview Communication Compliance features for Copilot and Insider Risk Management

EricaToelle's avatar
EricaToelle
Icon for Microsoft rankMicrosoft
Nov 19, 2024

Use of GenAI is quickly becoming the norm for organizations in all industries and of all sizes. As adoption continues to accelerate, GenAI is quickly transforming how organizations operate. However, this rapid adoption comes with its own challenges for security and compliance. Security and risk leaders are concerned about emerging risks, such as prompt injection attacks, as well as analyzing GenAI use for risks such as non-compliant or unethical use.

We are excited to announce new innovations in Microsoft Purview Communication Compliance to help mitigate GenAI risks. These updates include expanded coverage for more GenAI apps, broader risk analysis, and alert triage usability enhancements. Additionally, there is now greater integration with Purview Insider Risk Management for more comprehensive risk identification. Here are the details of these updates.

New ways to secure and govern GenAI interactions

Communication Compliance previously announced support for Microsoft 365 (M365) Copilot applications, including Copilots in Office applications, Teams, and other Microsoft 365 apps. To recap, organizations can analyze M365 Copilot prompts and responses (interactions) to detect potential risks, such as the presence of sensitive information, matches for content safety classifiers, or other keywords and conditions you specify.

Today, we are announcing an extension of these core Communication Compliance capabilities to cover more Generative AI (GenAI) applications, now in public preview. Communication Compliance can now provide analysis of user prompts and AI-generated output for non-compliant or risky usage for the following applications.

  • Security Copilot
  • Fabric Copilot
  • Microsoft Copilot Studio
  • ChatGPT Enterprise

Learn more about Communication Compliance capabilities for GenAI applications. 

Figure 1: new Communication Compliance Generative AI channels

The release of Microsoft 365 Copilot to the US government GCC cloud is currently planned for Q4 2024. We are happy to announce that Purview Communication Compliance will secure and govern Microsoft 365 Copilot interactions (prompts and responses) from the day of release in GCC. Communication Compliance can analyze interactions entered in Microsoft 365 Copilot and Microsoft Copilot to detect inappropriate or risky interactions or sharing of confidential information. It supports Copilot apps such as Teams, Outlook, Word, PowerPoint and more. Learn more about Communication Compliance for Microsoft 365 Copilot capabilities and supported Copilot apps.

GenAI introduces new security and safety risks that necessitate the implementation of additional controls. For example, malicious users may execute prompt injection attacks to induce unauthorized behaviors from GenAI, and users might generate content that infringes intellectual property laws. Today, we are launching new GenAI risk detection features in Microsoft Purview Communication Compliance, developed through the Prompt Shield and Protected Material classifiers from the Azure AI Content Safety team. These features enable the detection of risks such as direct and indirect prompt injections, and the identification of protected material usage in Copilot responses. This includes content sources like news articles, lyrics, code from known GitHub repositories, and software libraries found within GenAI-generated responses. Administrators with appropriate permissions can receive alerts and investigate potential incidents, thus facilitating more secure and compliant use of Copilot. Learn more about detecting prompt injections and protected materials.

Figure 2: new prompt shields and protected materials classifiers.Figure 3: Communication Compliance detects a potential prompt injection attack.

Admins want to easily see GenAI risks and mitigation policies in a single location. Today, we announced Data Security Posture Management (DSPM) for AI, now generally available. As a part of this single location for GenAI security and compliance management, we are expanding the policies section to include Communication Compliance policies related to GenAI. You can now view these policies alongside other GenAI policies across Microsoft Purview. Learn more.

Figure 4: Communication Compliance policies in Data Security Posture Management for AI.

New Insider Risk Management integrations

New Communication Compliance policy indicators are now available in public preview for use in Microsoft Purview Insider Risk Management policies. Insider Risk Management uses a full breadth of service and 3rd-party indicators to help you quickly identify, triage, and act on potentially risky activity. This new feature enhances security by adding a new option to share a signal with Insider Risk Management when a user sends sensitive information through communication channels, such as Microsoft 365 Copilot, Microsoft Teams, and Exchange email. The indicators can also share a signal when it detects risky Generative AI interactions such as prompt injections and use of protected materials. These new Communication Compliance policy indicators help to provide a more complete view across potential risks in your organization. Learn more about Communication compliance policy indicators in Insider Risk Management.

Figure 5: Communication Compliance indicators in Insider Risk Management.Figure 6: GenAI jailbreak alert from Communication Compliance in Insider Risk Management.

Now generally available, enrich the Communication Compliance triage experience with a user's risky activities and risk severity from Insider risk management, further empowering Communication Compliance investigators to take informed remediation actions. Enriching Communication Compliance with Insider Risk Management (IRM) context provides additional data on a user's historical risky activities and the alert risk severity that an investigator can use to inform their remediation actions. This enhancement empowers Communication Compliance investigators with comprehensive information about user behavior. By leveraging IRM's detailed risk analysis, Communication Compliance can better prioritize and address potential threats, ensuring a proactive approach to managing insider risks and maintaining a secure and compliant communication environment. This integration streamlines the triage process, improving the overall effectiveness of compliance investigations. Learn more about the enriched Communication Compliance triage experience .

Figure 7: The Communication Compliance triage experience enriched with Insider Risk Management context.

New policy settings and triage experience improvements

Communication Compliance introduces the ability to identify and flag potential workplace safety concerns in user-reported Teams messages, now generally available. Investigators now have an extra indication to decide what risk exists in user-reported messages and can configure an Inappropriate Content policy to proactively identify similar inappropriate content moving forward. Learn more about identifying workplace safety concerns in user reported messages .

Figure 8: Content safety risk detection in user reported messages.

Communication Compliance now notifies investigators about hidden content matches in Communication Compliance for a more efficient triage experience, currently in public preview. This feature is designed to enhance transparency by introducing a notification banner that alerts users when keywords are embedded in hidden content, such as hyperlinks or encoded strings, triggering a compliance policy match. With this update, investigators will be able to quickly recognize that some or all keyword matches are not immediately visible in the main content body, saving time and ensuring a more efficient triage experience. Learn more about hidden content matches in Communication Compliance .

Figure 9: The Communication Compliance triage experience notifying the admin that a hidden match was detected in the message.

Get started with Communication Compliance

Please let us know what you think about these features in the comments below!

Here are some resources to learn more about Microsoft Purview Communication Compliance.

Updated Nov 20, 2024
Version 2.0
No CommentsBe the first to comment