Microsoft is using Word and Excel user data for AI training by default, and opting out isn’t easy (Updated)
A hot potato: The controversial issue of companies training their AI models on user-generated content is once again under the spotlight. This time, the programs in question are the widely used Microsoft Office and Excel. The data gathering is enabled by default, and opting out is a laborious, multi-step process.
Update: Microsoft has reached out to confirm that it does not use customer data to train its large language models.
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
Microsoft’s connected experiences in Office analyzes user content to provide the likes of design recommendations, editing suggestions, data insights, and similar features.
On X, nixCraft pointed out that the Redmond firm has recently enabled a feature that scrapes users’ Word and Excel documents to train its internal AI systems. It’s turned on by default, too.
Heads up: Microsoft Office, like many companies in recent months, has slyly turned on an “opt-out” feature that scrapes your Word and Excel documents to train its internal AI systems. This setting is turned on by default, and you have to manually uncheck a box in order to opt… pic.twitter.com/wUfhBjcMOR
– nixCraft 🐧 (@nixcraft) November 24, 2024
As is so often the case when a company wants its customers to keep something enabled, opting out of the data collection is far from quick and simple. On Windows, it requires going to File > Options > Trust Center > Trust Center Settings > Privacy Options > Privacy Settings > Optional Connected Experiences and unchecking the box. Furthermore, once you untick the box, a prompt appears warning that disabling the option means some experiences won’t be available.
For those few who think this isn’t a big deal, Tom’s Hardware notes that a clause in Microsoft’s Services Agreement grants the company a worldwide and royalty-free intellectual property license to use your content.
“To the extent necessary to provide the Services to you and others, to protect you and the Services, and to improve Microsoft products and services, you grant to Microsoft a worldwide and royalty-free intellectual property license to use Your Content, for example, to make copies of, retain, transmit, reformat, display, and distribute via communication tools Your Content on the Services,” according to the clause.
Microsoft isn’t the only one guilty of this sort of underhand behavior, of course. Meta also uses public posts, comments, photos, and interactions with chatbots from Facebook, Instagram, Threads, and WhatsApp to train its AI models. Unlike in the EU and UK, those in the US do not have a straightforward way to opt-out – setting your account to private helps, but it still isn’t a guarantee this won’t happen.
In August, it was revealed that Nvidia, the company whose hardware powers the generative AI revolution, had been downloading 80 years of videos daily from YouTube, Netflix, and other platforms to train its AI models.
Microsoft hasn’t commented on the story. It’s possible that the outcry could lead to a clarification of its terms of use, much like Adobe did after after a popup suggested that the company could access and claim ownership of content made with its creative suite to train AI models.
