Microsoft has firmly rejected recent allegations claiming the company uses customer data from its Microsoft 365 applications to train artificial intelligence models. The controversy emerged after social media users raised concerns about Microsoft's data collection through its "Connected Experiences" feature and its potential role in AI development.
The debate centered around a privacy setting in Microsoft Office that enables "Connected Experiences," which allows features like online image searches and cloud-based services. This setting, enabled by default, sparked speculation about data usage for AI training purposes.
Microsoft's spokesperson directly addressed these claims, stating: "Microsoft does not use customer data from Microsoft 365 consumer and commercial applications to train foundational large language models." The company clarified that Connected Experiences solely enables basic functionality like co-authoring and cloud storage.
Frank Shaw, Microsoft's communications head, reinforced this position on social media, explaining that the setting only facilitates internet-dependent features and has no connection to AI model training.
The confusion was partially fueled by a Microsoft learning document from October 2024 that listed various Connected Experiences features that "analyze content" without explicitly stating exclusions regarding Large Language Model (LLM) training.
This incident reflects growing public sensitivity around data privacy and AI development. Similar concerns recently led Adobe to revise its terms of service, specifically addressing how user data relates to AI model training.
Microsoft maintains that while some customers may consent to data usage in specific scenarios, such as custom model development when explicitly requested, the company does not use general Microsoft 365 user data for training its AI systems.