Microsoft Purview compliance portal: Communication Compliance – Detect prompt injection attacks and mentions of protected materials in GenAI prompts and responses?

The Microsoft Purview compliance portal is enabling Communication Compliance, which includes the ability to detect risky Generative AI interactions using Azure Content Safety's Prompt shields and Protected materials classifiers. This feature helps organizations identify the risk of jailbreak prompt exploitation by malicious users and detect if GenAI responses contain branded or copyrighted materials to promote content originality and safeguard brand reputation. Communication Compliance is designed to assist organizations in detecting regulatory compliance violations, such as confidential information, offensive language, and sharing of adult content. The product is built with privacy by design and equipped with features such as pseudonymized usernames, role-based access controls, and audit logs to ensure user-level privacy. Communication Compliance is globally available for web cloud instances, and the release date is November CY2024, with a preview available in October CY2024.

Overall, this feature provides a new level of protection against potential risks in Generative AI interactions that can harm organizations. Its features help in identifying potential risks and maintaining regulatory compliance while preserving user privacy.

The post Microsoft Purview compliance portal: Communication Compliance – Detect prompt injection attacks and mentions of protected materials in GenAI prompts and responses? appeared first on M365 Admin.