Legality and Morality

Legal, Moral, & Ethical Considerations

Is there a legal requirement to report classified abuse/harassment to regional and national authorities. If so, at what scope?
If not a legal requirement, is there moral and ethical obligations to report abuse?

Should users have the option to view classified abuse? This will likely have to be a yes in order for users to help feedback into the model to improve predictions, at least at first. How can users be incentivised to avoid seeing the abuse (removing the purpose of the classifier in the first place)? Perhaps it can be made so that all abuse can be sent back and trained by a team of humans using tools such as prodi.gy (etc.) and therefore doesn't have to be viewed by the receivers of the abuse? Their incentive is to trust a team tackling this problem, whilst still providing them with the right to view content should they choose. This requires thought.

Does a browser extension / mobile app (etc.) breach the Terms of Service of certain sites (i.e., the scanning and analysis of potentially personal and/or sensitive information)? Perhaps having an opt-in user agreement when downloading the extension/app can circumnavigate this issue.
Example: Twitter and Facebook do not allow the scraping of content from the site (without express written permission). If data is being scraped for a model, it is in breach of the Terms of Service. This of course is not the case when using the API (as it is not scraping), but using an API for every site (Facebook, Reddit, etc.) is not a scalable solution (and perhaps not even feasible). Perhaps written permission needs to be granted. How have apps/extensions like Tune circumnavigated this problem?

How do regional, national, and international law differ on all of the above comments?
Is it the case that the 'product' only has to abide by local regulations?