Preventing Data Loss to Generative AI at the Endpoint

Microsoft Copilot, ChatGPT, Claude, and other generative AI tools are now embedded into enterprise workflows. Employees use them to summarize contracts, analyze financial data, debug code, and draft communications, often by copying, pasting, or uploading sensitive information directly from their computers.

This creates a new compliance risk. Sensitive data can leave secure repositories and enter unauthorized large language models (LLMs) in seconds, bypassing traditional network controls. Security teams may have AI usage policies, but they lack enforcement and defensible logs proving that sensitive data was protected.

In this technical session, Jeremy Moskowitz demonstrates how to prevent AI-driven data loss at the point of origin: the endpoint.

You will see how to:

• Detect and block sensitive content at the exact moment it is sent to an LLM
• Monitor and control copy-paste and file uploads into browser-based and desktop AI applications
• Apply controls based on data sensitivity, user identity, and device context
• Enforce protection consistently across Windows and macOS endpoints
• Generate tamper-resistant, audit-ready logs tied to user activity and AI interactions

AI governance without enforcement creates regulatory exposure. This session shows how to prevent sensitive data from being shared with generative AI systems, and how to prove that protection is working.