New Claims Raise Eyebrows On Google’s Gemini Accessing Personal Data Without Consent (Updated: Google Statement)

In the last few years, the trend of generative AI has picked up the pace.

From OpenAI to Microsoft and Google, the list of AI tools and chatbots keeps increasing. This has also given rise to a plethora of opportunities that come with increased usage but at the same time, the drawbacks related to privacy must be addressed.

Keeping that thought in mind, many are now raising questions about Android maker Google’s Gemini chatbot. The latter was accused of accessing and reading personalized documents related to tax via the tool’s latest sidebar.

Image: Solen Feyissa / Unsplash

This new sidebar was launched on an array of Google apps including Docs and Drive, giving the AI chatbot the chance to see what users are working on and at the same time, provide suggestions and an analysis of the documents at hand.

However, one user was baffled to see the tool go way beyond its authorized limits by reading his personal tax documentation when no consent was given. He explained how his tax returns were very confidential and for Gemini to summarize it was an eye-opener.

Seeing Gemini ingesting data present inside private documents is a huge wake-up call as it’s doing tasks that nobody has asked it to do and users aren’t aware of how they can stop this.

The fact that AI tools are pulling out commands without receiving any prompts is worrisome and now they’re racing to ask Google how they can disable settings for Gemini to stop it from breaching their privacy.

What was even more shocking is when the user actually found the setting to switch such actions off, he was confused as they were already disabled so why was the AI tool doing something that it was never instructed to do?

While tech giant Google does provide support documents that detail more about how Gemini should be used across Google Drive, the company fails in terms of detailing how the feature could be disabled or how to stop the AI chatbot from getting access to data inside Google Drive.

Google has failed to address the concerns of many who don’t think this should ever happen. It’s a huge eye-opener to the world in terms of what Gemini is capable of and how AI cannot be trusted at any given point in time.

Update, 16th July 2024.

Google's spokesperson explained that, “Our generative AI features are designed to give users choice and keep them in control of their data. Using Gemini in Google Workspace requires a user to proactively enable it, and when they do their content is used in a privacy-preserving manner to generate useful responses to their prompts, but is not otherwise stored without permission.”

Adding further:

"Your data stays in Workspace. We do not use your Workspace data to train or improve the underlying generative AI and large language models that power Bard [now Gemini], Search, and other systems outside of Workspace without permission.

There are several inaccuracies in this thread, particularly the notion that any data ingestion is occurring. It’s not. If the feature is enabled, the content from an open doc can be used to generate a summary in real time, but neither that summary nor the doc itself is stored. To be very clear all these interactions are ephemeral and no data is stored. 

In this instance, it sounds like the user was previously using Gemini via the side panel in Google Drive for PDFs and wished to no longer do so, which can simply be done by closing the side panel. 

The summaries referenced are completely ephemeral and do not change anything about how Gemini works either for that user or other users.

In terms of how it works: once you start using Gemini, we have settings that respect user preferences and carry over to subsequent uses of our products."

Read next: AI Writing in Education: Study Shows Alarming Detection Gaps
Tags:

Dr. Hura Anwar

A dental surgeon by profession and a freelance content writer by passion

You might like

View all
Previous Post Next Post
Share to other apps
Copy Post Link