Microsoft has released a new collection of tools and a guide to fix security issues that have arisen around the way the company’s generative AI (genAI) Copilot assistant handles information. Namely, the tool’s indexing of an organization’s internal data can lead to the AI assistant sharing sensitive information when it shouldn’t.
A Microsoft employee familiar with customer complaints tells Business Insider: “Now, when Joe Blow logs into an account and starts Copilot, they can see everything. All of a sudden, Joe Blow can see the CEO’s email.”
Business Insider reports that the behavior prompted several organizations to delay using Copilot for security reasons. “Many data governance challenges associated with AI were not caused by AI’s arrival,” a Microsoft spokesperson told the publication.
Instead, according to the spokesperson, AI tools like Copilot highlight how companies need to take proactive responsibility for how they manage internal documents and other information.
