Microsoft said it does not use customer data from office applications to train large language models. This happened after users began to suspect the company of the opposite.
The confusion stems from the privacy settings in Microsoft Office, which contain an option for “additional plug-in features.” It is enabled by default and allows you to search for images on the Internet and “find information on the Internet.”
Some users mistakenly thought that this clause also allowed Microsoft to train its artificial intelligence on user data, although this was not written anywhere. They called for privacy settings to be turned off so that Micorosft could not access the information. As a result, Microsoft had to explain Xthat it does not use user data to train AI.
In Microsoft 365 Apps, we don’t use customer data to train LLMs (Large Language Models). This setting only enables features that require internet access, such as co-editing a document.
Microsoft
The situation is reminiscent of the Adobe case, when in early 2024 users suspected the company of using their data to train AI due to unclear language in the terms of use. Then Adobe also had to publicly state that user data is not used for these purposes. (The Verge)
Source: www.iphones.ru
