When we upload our assets and eventually get them approved and then they enter the catalogues. Users and buyers are at the mercy of "luck" whatever luck is. A.i./algorithms etc.

But the reality is quite brutal and this is a simple example. If you search for something A.i. isn't very good at - insects and choose one that is quite common and select video you would expect to get examples of that insect (on this ocassion).

Each page has 100 results and in this example you have 4 pages of video results for - Scorpion
Fly. How many pages would you be willing to search through to get the right footage?

Beacuse page 1 of 4 has .... 66% of the results which are not even insects. Birds, fish, food, and dragons. Many are insects which arent a scorpion fly

It is tragic - https://streamable.com/p5rhgo

And that was just the video results.

Sky News today aired a story about gathering momentum from creative artists to call a halt to ilegal  A.I. data set usage because work is being plucked direct off their private commercial websites.

 https://www.google.com/amp/s/news.sky.com/story/amp/ai-art-generators-face-backlash-from-artists-but-could-they-unlock-creative-potential-12857072

Someone should pay attention or its gonna be expensive down the line. The US copywrite office decided in 2022 that A.I. generated work wasn't eligible for copywrite and began defending a law suit against a company that filed a law suit claiming this decision was wrong. Never the less at present A.I. generated work isn't copywritten. So anyone who wants to use it can do so free of charge.

They were open to exploring a change in the law if humans were involved. For instance if an artist or photographer used A.I. to enhance work. But the problem exists on what percentage of A.I. involvement does the copywrite cease to apply. They state they will look at the situation this year. Which explains the gold rush to get it up and running.

The UK government are now looking at the illegal use of artists (creators) via data sets which they are aware have been used illegally after pressure from many trade bodies. And will be looking at changing the law. But they state that initially this change in law will possibly require voluntary registration. But the big boys will rush to join and those who dig their heals in will pay the price. Because that's how it always works no matter what it is.

So Adobe ... look faster at a compensation model.

A paper by researchgate in January 2022 examined the potential for legal action against companies that use copywritten work to train their A.I.

Using Generative Adversarial Networks (GAN) to create images from data sets, which is how many A.I. generators work, they looked at the legal position regarding copywrite law, although newer methods exist such as CANs (Creative Adversarial Networks). The former would focus on the predominant features of a data set. If it was trained on animals it created animal like images even if clouds that contained animal like features were added. CANs were created to remove human input in the creation process but never the less using human data sets.

After examining copywrite law in this framework they state that Copywrite law may be triggered if the origin data was copywritten work. It varies from state to state but generally this fact is a given via reproduction laws. Even partial use is covered and requires the authorisation of the author/artist/creator ... this was never given. Implied authorisation by use of the site isn't enough because you have to be aware your images are being used for this purpose. 

Opinion - If the output of the A.I. isn't as yet copywritten then anything it creates is free to use and requires no payment.

If people are paying for those created images it proves that that they believe their purchase is protected by use laws. In this regard they believe their purchase is copywritten and safe.  And no doubt the Ts&Cs will assert as much therefore a company selling these pictures must be offering to protect the images because they are charging for them. This infers the sold A.I. output is coyywritten and protected by Adobe's legal framework.

If this is true they must have paid for the license to use copywritten work to create protected work which they sell. If they didn't they must give the A.I. generated work away for free. They can't claim fair use because they sell world wide and are also governed by TDM (Text and Data Mining) in Europe. Which excludes commercial gain.

You can make your own opinions but compensation now will be much cheaper than compensation later because now you pay for what you have used. Later you have to pay everyone because you won't be able to prove who's work you did or didn't use.

Researchgate source - https://www.researchgate.net/publication/357685384_Protection_of_AI_generated_photographs_under_copyright_law_pre_print_version_09_01_2022

Edit: I've just seen a payment today from POND 5 for data set use of my work. SS have also paid for use of my work. As we can clearly see ... regardless of the laws in place at present ... stock agencies do not want to be the last one digging their heals in.





 

Quick question
I've not sold anything on Pond 5 before but I sold a video. Yay for me. Had an email saying what I got. Went to my dashboard and looked at total earnings but this is less than the page that has all tha data per photo. So that says 40 ish dollars but total earnings is less. Normal? Not normal?