Google Gemini is apparently scanning private PDFs in users’ Google Drive accounts, the latest issue involving AI and privacy.

The issue was discovered by Kevin Bankston, Senior Advisor on AI Governance at the Center for Democracy & Technology. Bankston first noticed the issue when he realized Gemini had summarized his tax return, with being asked to.

Just pulled up my tax return in @Google Docs–and unbidden, Gemini summarized it. So…Gemini is automatically ingesting even the private docs I open in Google Docs? WTF, guys. I didn’t ask for this. Now I have to go find new settings I was never told about to turn this crap off.

Kevin Bankson (@KevinBankston) | July 10, 2024

In a long X thread, Bankston continued investigating what was happening, why Gemini had accessed his tax return, and how to turn the feature off. Needless to say, his experience is not reassuring. After searching and failing to find a setting that would prevent Gemini from accessing his private documents, Bankston asked Gemini to point him in the right direction—only to find that neither of Gemini’s suggestions worked.

Ultimately another user was able to point Bankston to Gemini Workspace privacy commitments that make it clear the AI doesn’t use inputted data for training. As Bankston points out, however, that still doesn’t settle the issue of being in control of one’s own data and prohibiting Gemini from accessing it all.

Bankston has some strong words for Google’s decision makers.

If this was a deliberate choice rather than some screwup, I can pretty clearly imagine the internal arguments that went into this very, very wrong @Google decision. So I’ll just say: listen a bit more to your privacy and responsibility people and a bit less to your growth people.

Kevin Bankston (@KevinBankston) | July 10, 2024

Bankston’s case is a damning indictment of Google—and the AI industry at large—and the reckless, irresponsible way such companies are addressing issues surrounding user privacy.

Google’s Response

Google reached out to WPN to emphasize that the company does not use Workspace data for training purposes, and no data is “ingested.” The company insists that any summaries are being done in real time, with no summaries or information from the document itself stored.

“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.” – a Google spokesperson 

There is still no word on whether the company plans to give users the ability to exclude specific content entirely, for those who want more granular control over what Gemini accesses—regardless of how that access occurs.