CourionAI
EN
Newsletter
← All news
google 2 min read

Google Is Rolling Out Controls to Stop Gemini Reading Files It Shouldn't

A new data loss prevention feature lets Google Workspace admins block Gemini from accessing sensitive Drive files based on content and labels, starting August 20.

A padlock layered over a stack of folders, guarded by a shield, with a small robot assistant reaching toward it

Google has started rolling out a feature called Gemini DLP, short for data loss prevention, to Google Workspace, the suite behind Gmail, Docs and Drive used by businesses and schools. The rollout began August 20 for accounts on Google’s faster “Rapid Release” track, with the rest following gradually starting September 1.

The feature lets Workspace administrators set rules that block Gemini, Google’s AI assistant, from accessing certain files in Google Drive, based on what’s in them or how they’re labeled. Think a spreadsheet marked “confidential,” or a document containing content that matches a company’s sensitivity rules: with Gemini DLP turned on, an admin can stop Gemini from reading or referencing that file at all, even if an employee asks it to. A related feature, DLP for Studio, extends similar protection to Studio flows, Google’s tool for building automated AI workflows, letting admins block or require human review of a flow based on where its data comes from and what it touches. For now, Gemini DLP only covers Google Drive, with support for other Workspace services expected later. It’s available to customers on Frontline, Enterprise, Education and Enterprise Essentials Plus plans.

What’s actually going on here: AI assistants like Gemini are genuinely useful specifically because they can dig through a company’s own documents to answer questions or draft new ones. But that same ability is exactly what makes IT and security teams nervous: an AI assistant with broad access to a shared drive can easily surface information that was never meant to leave a specific team, project or client file. Data loss prevention tools have existed for years to stop that kind of accidental leak in email and file sharing; Gemini DLP is Google extending the same idea to AI itself, treating the assistant as just another channel that needs the same guardrails as everything else. It’s also a response to a real business hesitation: many companies have held back on giving AI tools full access to internal data specifically because they lacked fine-grained controls like this one.

What this means for you: if you’re an employee using Gemini at work, you likely won’t notice anything different unless your company sets up new restrictions, in which case Gemini may simply decline to reference certain files it used to be able to see. If you’re an IT admin or work in security, this is worth checking out directly, since it addresses a real gap that’s been a genuine blocker for wider AI adoption inside cautious organizations. For everyone else, it’s a small but useful sign that AI tools are slowly catching up to the access-control standards businesses already expect from the rest of their software.

Sources

Source: https://workspace.google.com/blog/ai-and-machine-learning/enterprise-security-controls-google-workspace-gemini

Next story

Hugging Face Is Reportedly Exploring a $13 Billion Sale

The open-source AI model hub has fielded acquisition offers that would nearly triple its 2023 valuation. No deal is confirmed, and its founders have long resisted a sale.

A smiling gear-and-circuit face held up by two hands above a small row of shopfronts