Several days ago, news broke out about Microsoft scrapping data in Office documents to train its AI models and making it very hard to opt out. While the general response was typically negative, some users noticed that the report was not quite accurate. It turned out they were right.
Microsoft has responded to allegations and clarified that it does not use user data from Office documents for AI training. "Optional connected experiences," the feature that was accused of data scrapping, does not relay your document data to Microsoft. It is only used for additional online capabilities in Office products, such as cloud fonts, templates, weather on calendar, co-authoring, and more.
Some users responded to the report with a link to the official documentation describing what "Connected experiences in Office" mean, and Microsoft confirmed that in its own reply on X (via The Verge):
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. https://t.co/o9DGn9QnHb
— Microsoft 365 (@Microsoft365) November 25, 2024
While the official documentation is pretty straightforward, it is easy to understand why some people got spooked. It says that connected experiences "analyze your content," and some falsely assumed that means AI data training. These days, with every tech company obsessed with AI to oblivion, it is not surprising to see people overreacting to the lack of transparency.
It is good to see Microsoft clarifying the situation and putting an end to fake news. However, it would be nice for the company to update its documentation so that it explicitly states that Office apps do not send your data for AI training. Adobe was forced to do that not so long ago, and it looks like it is now Microsoft"s turn.