Seriously, an open source version would be awesome. You could connect it to your own server running whatever local models you want without needing to worry about that audio/video being processed by some large corporation willing to sell you out along with your data.

It doesn’t break the law at all. The courts have already ruled that copyrighted material can be fed into AI/ML models for training:

https://towardsdatascience.com/the-most-important-supreme-court-decision-for-data-science-and-machine-learning-44cfc1c1bcaf

This has nothing to do with RCS from what I read on the article. It looks like the UK wants to be able to tell companies to disable security features such as End to End Encryption so that they can view the messages.