What is the limit of transformative works here?

If I build an auto encoder neural network and train it to learn and return content exactly, that’s probably not allowed, but if I train it to rephrase the news, is that allowed?

Doubtful. It would be the same as quoting an expert in general terms without giving a proper citation (see APA paraphrasing).

The worst thing anyone could do about this technology is further empower copyright.

Knowing facts seems hard to copyright.

But at the same time, you could not run a newspaper through a thesaurus, word by word, and claim that’s a wholly original publication.

The general questions for protection seem to be: is the use minimal, and does it serve the same function? One news site copying another is not okay. But a comedy site riffing on the news, especially on that news site in particular, is clearly distinct. And one news site mentioning the conclusion of another news site’s article seems fine, even if they do not explicitly reference the source. They would tend to do so for the sake of building audience trust… but they could just treat it as unsurprising.

If you reproduce long-form content verbatim, it doesn’t really matter how many hoops you jumped through to get there.

Building a chatbot using the entire corpus of available English text doesn’t sound minimal - but every individual work makes a minuscule contribution. If a one-gigabyte model was trained on one million books, the takeaway from each book is about the size of this comment.

It argued in a court filing that it’s not located in Ontario and does not do business in the province.

OpenAI also argued the Copyright Act doesn’t apply outside of Canada.

The DST needs to come back immediately, at a higher rate and lower threshold. These US tech companies have zero respect for democracy and the rule of law.

Brazil knows what’s up, at least, and isn’t taking this stuff lying down