Is Microsoft really using your Word & Excel documents to train its AI?

Microsoft has now refuted the viral online claims that accuse the Redmond company of automatically scraping users’ data from Word and Excel to train its AI models.

The confusion arose from a default privacy setting in Microsoft Office that enables “optional connected experiences,” which allows users to search for online content or co-author documents. Some, however, say that it’s used to train large language models (LLMs).

“If you are a writer who uses MS Word to write any proprietary content (blog posts, novels, or any work you intend to protect with copyright and/or sell), you’re going to want to turn this feature off immediately,” the now-refuted claim reads.

Microsoft has a default feature in Office 365 that automatically scrapes and processes data from Word and Excel to enable “connected experiences,” which lets users easily collaborate through design recommendations and data insights.

“In the M365 apps, we do not use customer data to train LLMs. This setting only enables features requiring internet access like co-authoring a document,” Microsoft clarifies in a reply to the viral tweet, which, at the time of publication, has garnered 3.6K likes & 1.8K retweets.

But, given Microsoft’s not-so-good track record with this issue (more specifically with the nightmarish PR disaster of the Recall feature), folks still do have the right to complain.

Recall is a staple feature for Copilot+ PCs, which is now rolling out for beta testing for Snapdragon-powered devices. The feature lets you recall anything you’ve done on your desktop by periodically taking snapshots of it, but some say that it could be a security loophole, if not done correctly.