Skip to content
Zenteck
Latest
AI Tools

Gemini Finally Lets You Keep Chat History Without Training AI

ByAlex Silva2 min read0 comments
google-separates-gemini-training-and-chat-history
Photo: ZenteckGoogle separates Gemini training and chat history

Listen to article0:00 / 0:30

As the report from Ars Technica details, using workplace tools powered by generative models often forces an uncomfortable choice between convenience and privacy. Previously, keeping a searchable log of past Gemini interactions meant letting Google use those exact inputs to improve future models.

What changes in your daily workflow?

According to Ars Technica, Google is rolling out independent toggles under the Gemini activity settings. Instead of an all-or-nothing approach, professionals can keep their searchable history active while explicitly opting out of model training for any future interactions.

However, this update comes with a catch regarding ephemeral data. When users disable history entirely, Google will extend the temporary retention window from 72 hours to two months. This adjustment is designed to help systems process feedback and detect abuse, but it means unlogged chats remain on servers much longer than before.

How to update your Gemini privacy settings

As the rollout reaches your account, securing your workflow requires a quick manual adjustment:

  • Open your Google account or the Gemini app settings.
  • Navigate to the Activity or Gemini Apps Activity panel.
  • Locate the newly independent toggles for model training and chat history.
  • Switch off the AI training toggle while keeping your history preferences intact.

Keep in mind that this opt-out only applies to future inputs. Anything previously uploaded, including text, attachments, or tool calls, remains part of past training sets and may be retained by Google for up to three years.

Verdict: Is it enough for sensitive work?

This update removes a major friction point for professionals who rely on past context to build projects. While historical training data cannot be purged retroactively, the new granular control makes Gemini a safer environment for handling ongoing tasks without feeding proprietary prompts back into public models.

Sources

  1. Google will let you opt out of AI training in Gemini without losing your history — arstechnica.com

Frequently asked questions

Can I remove data already used for Gemini training?
No. According to Ars Technica, opting out only applies to future interactions. Data previously fed into Gemini remains part of existing training sets.
Where are the new privacy toggles located?
They are located under the Activity section in your Gemini settings, or via your Google account under Data and Privacy and Gemini Apps Activity.

AI Briefing in 3 Minutes

Get what actually matters in AI & tech delivered free to your inbox. Zero spam, 1-click unsubscribe.

By subscribing, you agree to our Privacy Policy.