In the context here, I believe they mean someone can _use_ the information externally. Yes, we rely on external vendors for business critical operations like email, or our cell phone providers for customer communications. That does not mean we’d usually grant them access to view, distribute or make use of in any way the content of those interactions without external governance requirements (court order, legal discovery, compliance/regulatory requirements).

For example in the US, a court could compel any external vendor we outsource our data to for operational, processing or storage purposes to access that data for any use it deems necessary. This was the expectation of any commercial business relationship across the country.

Gmail broke this with ad data, but the data was meta-data - information about information, like ad groups and categories relevant to your org.

Now publicly published content is being ingested and used to train models, at this stage I would assume if you share Docs publicly that would be in scope.

Proton has a good article explaining what is and isn’t ingested, and what they have the right to use: https://proton.me/blog/google-docs-ai-scraping