Perhaps reach out to Adobe? Or mat here - and see if he can pass the feedback onto the devs?

While I agree that the submission requirements do say it shouldn't have identifiable locations/people/etc - so in this case - I would say this should have not gone through according to the current specs...

However, making a reason to reject based on something that could be construed/potentially "offend" someone - while yes, of course tragic/etc, and definitely very tragic for anyone personally affected - it would not be right to reject something because someone could be potentially offended, otherwise, where do you stop? Slippery slope. Some people might not like pictures of cows (i.e., vegans, east indians, etc). Some people might not like pictures of churches, or conversly mosques, (i.e., they "feel offended" by a particular religion), etc. Some might not like political parties, political stances, etc. Some people might not like 'black' people, others might not like 'white' people, others might not like 'asian' people, etc, etc. So 'feeling offended' is not a good reason to reject. (Total aside - last 3 years should have been an eye opener for many in terms of what really happened then, as well as what really happened WWII/etc - people are deliberately having their emotions/thoughts manipulated - but entirely different topic).

Since we have no bargaining power, what should be in an ideal world is irrelevant. We simply get a one-time payment for using our copyrighted works to train AI, but all profits from AI then go to them. Legally, it's probably legal. We weren't asked if we agreed with use or the amount of the one-time payment, but that's obviously not a legal issue. We're just screwed and it's only a matter of time before we get kicked out and fully replaced. And this also applies to those who generate AI images, it makes no difference in principle, it's just a temporary fill in for the demand for AI images.

Actually, you have an incredibly amount of power, but you need to use it. Everything from initially contacting the agencies politely, to contacting/discussing with a lawyer, to class action, etc, etc. "They" would like you to think you have "no power", but the contrary is true. You just need to realize that. If you have an attitude of you've lost - then you have already - so you need to realize that and start taking action NOW.

Why not contact iStock and show them how easy it is to replace - perhaps they will come up with a better watermark that can't simply be replaced with generative fill, and/or cropping it?

Interesting, thanks for sharing... now - over what period were those best sellers? I noticed you mentioned one of your pics was from 2014... does that mean these are your 'all time' best sellers since you've been doing stock?

once again, you neither understand how these generators work, nor the massive programming involved. have you worked on such huge projects?

if everyone can optout at any time the training would have to be continuous, and there's no indication original images used would still be available - where are those billions going to be fud and how would they be able to identify your work?

but again, you dont understand how these work -; once trained, there is NO way to trace back to original training set.

Once again, you are mistaken, and it sounds like you have limited programming knowledge. Is that the case? I actually do know what I am talking about. And yes, I actually have worked on big dataset projects, so I know EXACTLY what I am talking about. It is NOT a "massive programming" undertaking. The "massiveness" is simply processing the data - but that is why there are now HUGE HUGE HUGE server farms - which make that a pretty simple task. (I.e., you've heard of google, right? They regularly refresh their "search engine database" (and actually archive a LOT more than simply archiving computer models of images)).

If a company is using an "out of the box solution" (aka a lazy man's way of doing things) - then yes, without ANY changes whatsoever, as far as I know - the 'out of the box' solutions don't have that pre-installed. HOWEVER... if you hire a couple programmers and tag the data - then yes, you CAN do that.

Answering your questions:
1. The dataset opt-out/opt-in would be at a set interval - and process them all in batch. (I.e., say you had 1000 people that 'opted-out' - it would not happen at once - it would be 'batched', i.e., say 8pm every day when the scraping/modelling was done - anyone who was 'opted in' would get processed, and those that 'opted out' would not).
b) In terms of the interval - depends on the server farm the company is using & the speed (not "really" an issue longterm - but short term because they are currently 'learning' - and depending if they are stealing 5 million images, or 5 billion - makes a difference). So could (for now) make the process say 1x every 3 days (to refresh the background modelling/dataset).

2. And again - you are mistaken - you actually CAN "tag" the data to be associated with the processed data. YES - it does (most likely) require revising the current algorithm - and YES - many programmers are lazy (or in some ways incomptent), so you actually need GOOD programmers (full stack would probably be best because they can see the 'bigger' picture usually) - but you CAN do it.

Let's take a hypothetical example (SUPER simplistic, but illustrates the point).
User A has pictures of chairs & lamps, i.e., "ID001".
User B has pictures of dogs, i.e., "ID002"
User C has landscapes, i.e., "ID003".
User D has landscapes, i.e., "ID004".

When initially 'modelling' the data, when the computer model for a "landscape" was put in, tags would be associated with (essentially what is a computer model/representation of the 'idea' of a 'landscape'). Likewise, for dogs/chairs/etc.

Then, say a prompt is "dog sitting in a chair".
User C/D's input was not used, so not relevant. However, A & B had the "computer model" accessed to "compose" that representation - so they would be "credited" with having composed that.

Then say it was "dog in landscape".
Now B/C/D were used, so would likewise be credited.

Say each image (for simplicity) was $1 to compose. And say it was a 50-50 split between the "ai tool/engine" & contributor.

In the first scenario, A&B get 25 cents each. In the second, B/C/D get 16.7 cents each.

THAT is how it would work.

It's possible the companies already designed their systems to be able to do that (for other reasons, I.e., logging, revising algorithms/tweaking "representations" of chairs (or "hands", "faces", etc)). If not - the algorithm CAN be redesigned.

It is simply a matter of doing it, & would require some thinking.

It is not "massive programming". The "massiveness" is simply processing the data - but again, almost an insignificant point because of the HUGE MASSIVE server farms - and how much "cheaper" it is becoming to process massive amounts of data.

It IS in fact a relatively simple thing to do. It is simply a matter of DOING it.

Yes, like other people said - if I need something in a series/different poses/etc.

I have an aerial image of a town shot with a wide angle lens. And in this photograph, there are several parked cars but they are very distant and impossible to read the license plate numbers. It is possible that a car expert may be able to identify some of the brands or models of the cars without zooming in though some people may struggle to do this. Would such an image be suitable for commercial usage in Adobe Stock?

There is also a brand name visible on a supermarket in the photo but it is barely visible due to the distance. I'm planning to clone the brand name out but that is probably overkill. You would be able to identify the brand name if you zoomed in to the photo.

I know in the past, stock agencies would normally accept photographs of cities etc for commercial usage if it was a wide shot (showing a city skyline etc and everything is distant.) However, these current warnings about logos etc on the AS submit page has made me extra cautious and I don't want to take any chances.

I could also add that there are some old historical B&W photographs displayed on the wall of the supermarket that are possibly in the public domain. Though once again, they are extremely distant and hard to identify as photographs. Though a local would probably be able to recognise them as photographs.

My experience is 'generally speaking' if they just look like "generic cars" - then usually it is fine for commercial use. If however - you say could easily identify the brand (i.e., say a row of lambourghinis), or easily identify license plates, street names, business establishments, etc - then yes, that would become editorial footage.

you're off by several orders of magnitude and show you haven't really studied how ai training works - there millions++ of datapoints (not '100 ') and the identity of the images used in training is lost in creation of the dataset as there is no longer a simple correspondence between initial pixels (24MP/image at minimum) and the resulting dataset.  so when an ai image is generated there's no way to trace back

dont know your programming bkgd, but the solution you present is enormously complicated

For simplicity - I phrased it that way, and used easy to understand numbers. I very much do know what I'm talking about, and yes - it is possible, actually quite simple to do.

Now it is true many programmers are lazy - and use existing algorithms, instead of "thinking" - but definitely very easy/doable. I'm not talking about corresponding "pixel images" - I'm talking about the "model/representation". It is very easy, and very possible to "trace back" source images - or more specifically "source data" that were used in composing an image.

You'd have to re-code a few things, and to do things retroactively - run it on previous data (compositions/queries) - but almost every single company (i.e., midjourney for example) tracks EXTENSIVELY with MASSIVE web logs/stats/etc. So very easy not only to process payments retroactively - but also going forward.

It is just a matter of doing it.

---------------

Hi Matt,

Thank-you for the FYI. While it is nice to compensate authors, and certainly appreciated -

(a) Doing it the way other companies did (i.e., "take first, ask permission later") does not feel right, and is not right.
(b) A more honest/ethical/equitable approach would be doing fractional payments (as opposed to a lump sum) for EVERY image generated with the tool going forward. I realize that is not what other companies are doing - BUT - it would certainly be a much fairer system. AND... adobe could be a leader here... If you wouldn't mind passing that on, that would be appreciated.

Basically -
(a) Opt-in/opt-out system. Contributors CHOOSE whether they want to participate. Works are added/removed from the training set, depending on the setting.
(b) For a 'fair' system (where you'd most likely get contributors WANTING to participate) - contributors benefit for EVERY SINGLE "AI" IMAGE generated.

How do you do that? It's quite simple, really.
- When the neural networks are set up - the ID # of the images is recorded for the data inputted - i.e., a "data point".
- When a customer "generates" an "AI" image - it "pulls" from sometimes tens, or hundreds or thousands of "data points" to create that image. All the ID#'s of images used in composing that "AI image" is recorded.
- Each contributor - image ID - is given a fractional portion of that generated sale. Which, obviously adds up the more images created.

Doing it this way is certainly much more ethical, AND equitable/fair - and most likely you'd have people WANTING to make images when they know they will be compensated for, not with a tool that is designed to "replace" them.

It also CAN (and SHOULD be) done retroactively - and is very easy to do so.

Going forward it is also very easy to do so.

So for example, let's say:

a) A customer pays $50 for an "AI" tool, and makes 500 images. So each "image" is worth $0.10.
b) Let's say one of those random images "used" 100 contributor files to do so in their neural network.
c) Using the current arrangement (33%), payment would work out as follows. (As an aside, the % should be upped significantly for contributors, because once the tool is in place, adobe doesn't really have to do much 'maintenance'. The 'work' is image creation. I'd suggest a 90% contributor/10% split, or at least 80% contributor-20% adobe. But a different topic).

But for now - using the 33% idea... $0.067 to adobe, $0.033 to the "contributor pool" for the image created.
100 'images' used to create the "ai" image, so $0.033/100 = $0.0033/contributor.

Obviously, for a single image that is not much - BUT - it also obviously quickly adds up, as 1000's of images are created with the "AI" tools.

Certainly much fairer, and equitable.

And OPT-IN/OPT-OUT respected. if a contributor chose to "opt-out" - then their data points would ALSO BE REMOVED from the dataset for future "AI" image generation. "OPTING IN" is likewise very easy - it simply 're-adds' the datapoints to the training set for "ai" image generation.

Programatically VERY EASY to do - although it requires a bit of work to set up. And doing it this way more likely to have contributors WANTING to participate, as opposed to getting very upset/annoyed because it was simply "taken" from them.

THAT is much more along the lines of "responsible AI", with creators in mind at the "center". Not the "pay once to you, we benefit forever on your works" model which creates resentment, and actually discourages future image creation (which long term, will make a useless "AI" tool, as it quickly becomes outdated).

It also makes Adobe a HUGE amount of money going forward, with nice consistent revenue for very little effort or work, and happy contributors that benefit too.

Thanks. everything seems sorted now.

I sent them a message yesterday lunch time and had a few e-mail exchanges to answer queries their engineers had and they had the problem fixed by 5 pm. Excellent contributor support.

What e-mail did you use? I've tried sending e-mails to what I think is the support address, but never yet gotten a response... Thanks!
PPS - have you made any sales from them? I haven't yet - but I do sell higher priced clips. Wondering if I should try changing the price point. Thanks!