No, Microsoft does not employ Office documents to train its AI

Microsoft says it doesn’t employ customer data from Microsoft 365 apps to train its artificial intelligence models. The explanation addresses reports circulating on the Internet IN last few weeks claiming that Microsoft required Word and Excel users to opt out of training the company’s artificial intelligence systems.

The confusion stemmed from a privacy setting in Microsoft Office that toggles “optional connected experiences” — a feature that Microsoft says helps users “search for photos on the web” or “find information available on the web.” This switch is on by default and there is no mention of AI training in the disclosure. Similarly, a Microsoft training document published on October 21, 2024 appears to have contributed to the confusion by describing a long list connected experiences in Office that “analyze your content” without explicitly excluding AI training under the Large Language Model (LLM).

“In M365 applications, we do not use customer data for LLM training,” he added Microsoft 365 X account saidresponding to claims. “This setting only enables features that require Internet access, such as document co-authoring.” Microsoft’s communications chief, Frank Shaw he also interfered with Bluesky to refute claims.

The Adobe and Microsoft incidents suggest that people are increasingly concerned about tech companies using their personal data to train artificial intelligence models without explicit consent. This is an understandable concern, given that companies like Meta, X, and Google default to AI training for their users, and that extensive amounts of online content are collected for this purpose.