Yep that would be another option. Should we have a standard format?
I mean what I find hard to implement as regulation is, for example, how do we know what data exactly is collected, and if we get a hairball of pointers what’s the use of that …
We have a hard time with getting a useable form of the content we create on platforms, especially for migration on other platforms/tools. I imagine if that is so hard, offering raw ML sets for every single tracked users is a whole lot harder.
Maybe a W3C of ML could work?