Microsoft Announce New Agents for Security Copilot and additional features for Securing AI and AI for Security

I assume everyone interested in Microsoft Security has seen yesterday’s announcement regarding the availability of new agents in Security Copilot.

I found two announcements that I found particularly interesting towards the end of the post.

Firstly, the release of some tools in Microsoft Entra to help mitigate against the risks associated with ‘Shadow AI’, a topic we discussed on our pod several weeks ago <ShamelessPlug>Search for #CloudyWithAChanceofInsights on your platform of choice.</ShamelessPlug>. This feature allows for more granular control of user access to GenAI tools. It should, therefore, decrease the risk associated with AI tools that are not officially supported and controlled by IT.

Secondly, and in part related to the first, is the ability to create DLP policies in Purvie for Microsoft Edge, which can be used to provide a level of control for users accessing these tools via the browser and will help mitigate data loss when using services including ChatGPT, Copilot Chat, DeepSeek and Gemini. Please note, that this feature is currently in preview.

Finally, and I know I said two, but it is nice to see some love given to Teams in the form of phishing protection for malicious URLs, including real-time detonation of attachments and links. This feature should be available in April