[sieabah]

If they had bothered to read the license agreement they would know whoever generates the art owns the copyright. Since it's a pay-for action the copyright is owned by the payee.

So really they generated these and never bothered to do the research of their own question.

[teddyh]

I have, like the article author presumably also does, a profund doubt as to whether generated works of this kind can be free of any copyright as long as the tool used is itself created using myriads of copyrighted works (as training data). I certainly do not trust the claims of the tool creators; they have all the incentive to ignore any copyright problems in order to get a tool which is usable.

And, as the article states:

> But seriously, how creative and original can you be with something that is trained on the works of millions of other creators?

> To me, it is unclear whether you can actually call these works your 'own' at all, because there's always someone else's touch on it.

> […] users of DALL-E will also never be sure whether they are generating something that is 'theirs' or just a knockoff of someone else's work.

[hedora]

OK, so I give you license to use this URL I just generated to generate your own stuff.

It's pay for action (send me a penny if you find anything worthwhile), and the copyright is owned by the payee:

https://images.google.com/

[meowkit]

The URL has not been "generated" in the same sense. You are retrieving an existing string. The images from google are not "generated" in the same sense, they are indexed from google's search algorithm.

The generative models, specifically for DALLE here, compute pixel unique images. You might say these models index a subset of an extremely high dimensional space (pixel count * RGB color values) using a query. Traditional search engines build an index from nothing and then use a search query to find the best matches in a more discrete space.

[teddyh]

If I made a website where you could type text and get images, and I said that your held the copyright to any images you got, could you safely act on that assumption? What if my web site was simply a proxy for Google image search?