Join our Newsletter — 33% off our NHI Course

Why are ChatGPT, Copilot and Gemini seeing sensitive uploads?

 

(@nhi-mgmt-group)
Member Moderator
Joined: 2 years ago
Posts: 20217
Topic starter  

TL;DR: 40% of file uploads to AI tools contained personal or payment data, according to Netwrix case work. Endpoint controls need to stop those uploads before regulated data leaves the device and reaches external AI services.

Editorial analysis by NHI Mgmt Group, based on content published by Netwrix: “Verhindern Sie, dass Ihre sensiblen Daten in der KI-Cloud landen – bevor es zu spät ist”.

Key questions

Q: When should organisations prioritise endpoint DLP over gateway inspection?

A: Prioritise endpoint DLP when employees regularly use browser-based AI tools, desktop assistants, or local AI applications.

Q: What breaks when employees use personal and corporate AI accounts interchangeably?

A: Interchangeable account use breaks attribution, policy enforcement, and data handling assumptions.

Practitioner guidance

  • Block sensitive uploads at the device boundary Use endpoint policy to inspect files, clipboard content, and drag-and-drop activity before users can move regulated data into AI tools or web apps.
  • Classify the AI channels employees actually use Map sanctioned and unsanctioned paths for ChatGPT, Copilot, Gemini, browser sessions, and personal accounts so policy covers the real workflow.
  • Enforce uniform rules across endpoints Apply consistent controls on Windows, Linux, and macOS so the same sensitive-data policy follows the user regardless of device type.

Bottom line: Sensitive information is reaching consumer and workplace AI tools through ordinary employee workflows, which creates a prevention problem at the endpoint boundary.

Glossary terms: Prompt-Boundary Governance, Shadow AI, Endpoint Data Loss Prevention (DLP), Managed AI Account

Explore further

View Full Forum →  |  NHI Foundation Course →  |  Our Services →  |  Read the full analysis →


This topic was modified 2 days ago by NHI Mgmt Group

   
Quote
(@mr-nhi)
Member Moderator
Joined: 6 months ago
Posts: 20707
 

AI cloud data loss prevention now sits at the endpoint, not only in the DLP stack. The article describes a user behaviour problem that classic perimeter-style DLP cannot reliably contain once data reaches consumer AI services. The practical shift is that policy has to evaluate the action before the upload or paste occurs, because after that point the organisation has already lost timing advantage.

A question worth separating out:

Q: How should security teams stop employees pasting sensitive data into AI prompts?

A: Security teams should control the browser or endpoint where the paste occurs, not rely only on network or file DLP. Classify text before it is submitted, block sensitive categories in consumer AI destinations, and log the event to an enterprise identity. The goal is to stop disclosure at the moment of composition.

👉 Read our full editorial: Preventing sensitive data from reaching AI clouds with endpoint controls



   
ReplyQuote
Share:

Free weekly newsletter

Subscribe to the NHI & AI Identity Journal

The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.

Bonus 33% off our NHI Course when you subscribe.