The end for original content creators. [Topic edited by moderator for clarity]

[NeoLeo]

I want to share the results of my case. What responses did I receive from Adobe?

I will briefly describe the details of my case:
I found that one author (actually there are many), takes images from my portfolio, uses them as a prompt in Midjourney, Stable Diffusion or Adobe firefly (it can be any AI generator, where there is the possibility of img2img) and generates similar AI works, and then sells these images on Adobe Stock. Also he copies the description of my work. "n the attachment, you will find a picture with several examples.

In Adobe's rules, I found this information:
Your Generative AI content prompts, titles or keywords:
-May not contain references to third party intellectual property

If I understand correctly, you can't use someone else's images as a prompt and img2img.

I sent the DMCA notice to Adobe's Intellectual Property Agent (copyright-stock AT adobe.com), reached out to Contributor Support, and communicated with the Adobe team on Discord.

These are the answers I got:
Adobe Intellectual Property Agent: Thank you for your message. We are unable to take action on your report. The allegedly infringing content appears not to be substantially similar to your content.

Contributor Support: The team who handles and deal with these issues is the one who already reviewed your claim and responded back > newbielink:mailto:copyright-stock@adobe.com [nonactive]

I am afraid we can't help you from here. You are more than welcome to send them another message and see if they can review your case again, or provide more details. But if they reviewed your claim and determined that these content do not infringes copyright or other IP rights, then I am afraid there's nothing else we can do. Your case has been reviewed, a decision has been made and a response has been sent back to you already.

Adobe team on Discord: How can you prove the member took your photos and created a file with it? it has similarities, but it is hard to prove. Yes, I see a resemblance, but it's not a direct copy. Ai can really bring our various images. unless your design is exactly/part copied, then you can make an IP claim.


I'm a little confused and have a few questions:
- Adobe themselves have established rules on uploading AI content, where it is forbidden to use other people's work as a prompt for img2img. However, the original content author has absolutely no means to prove that their picture was used in img2img.

- Adobe should be aware of how AI works. It is not possible to generate such a similar picture to the original using only text prompts. Obviously, an image was used as the prompt here as well. For example, try writing a prompt in Midjourney or Adobe Firefly like "a young girl in a yellow hoodie and red pants riding a scooter" and compare the resulting image to mine. You won't get an image as similar as mine without using my image as a prompt for img2img."


We've received an official response from Adobe:
Anyone can take any picture from any portfolio, upload it to Midjourney, and use it as a prompt, then sell it on Adobe Stock. The original author will have no means to prove that img2img was utilized. Adobe doesn't recognize any infringement, despite its clear presence to me.
This situation significantly disadvantages original content creators, and it's not entirely clear how to proceed."


@MatHayward, I would like to hear your opinion on this matter. You've conducted podcasts and promoted AI content on Adobe in this manner. You've spoken extensively about the benefits of AI, but you haven't addressed the challenges and safeguards for original content creators. We continue to see millions of accepted images with six fingers, three legs, and other anomalies. Would you be interested in hosting a new podcast where you address issues like this? How does Adobe protect authors from img2img?"

I would like to hear the community's opinion. What are your thoughts on my situation? I may be mistaken in some aspects, and I would appreciate it if you could point out any errors in my reasoning.

Thank you for your attention.

[Moderator moved the thread to the correct forum]

[Abambo]

Are your assets generative AI?

[NeoLeo]

No. I Don't use AI at all.

[Abambo]

IMHO, the assets from the other party are pretty close to yours.

[Nancy OShea]

Nobody can claim ownership of geometric shapes or ubiquitous avatars of doctors, hands or girls on a scooter.  That's not protectable. Sorry.

The styles & themes may be similar BUT they are not, not, not the same.  Obvious differences can be seen in scale, geometry, shading and details between the two sets of images.  I don't think you have a leg to stand on because 'similar' is not legal infringement.

I wish I had better news for you but I call things as I see them.

[Abambo]

Nobody can claim ownership of geometric shapes or ubiquitous avatars of doctors, hands or girls on a scooter.  That's not protectable. Sorry.

By @Nancy OShea

Apple claimed ownership of a round button…and rounded edges.

There is obviously a pattern of similar assets, and that is highly suspicious. I'm absolutely not sure how a court would handle this.

As for Adobe, they are masters in their house.

Some assets have (even in this small size) quality issues. But that's a different theme.

[TatyM]

Hello, in my case they copy detailed description of my photo and generate IA from my original text. Then they publish IA image with my description. Adobe do not care about this. Is there sence to make good detailed description for photos? May be few words is much better?

[Jill_C]

Titles must fully describe the asset. If you shorten the titles, omitting important information, you will drastically lessen the chances of a Buyer finding your asset.

[TatyM]

I understand this but they copy full description and make AI image from the text that I write for my foto.

[Abambo]

I understand this but they copy full description and make AI image from the text that I write for my foto.

By @TatyM

There are not 26 choices.

[Jill_C]

You do not own a copyright to the words that you use in your image titles; and in fact, using those words as a text prompt doesn't guarantee that an "imitator" would be able to generate the same image anyway. As an experiment, I clicked on one of the recent accounts highlighted on the Insights tab and used a few of the titles, word for word, as a text prompt. The images that were generated was nothing like the original image, which only proves that titles aren't the same as text prompts. Just focus on creating original content and add appropriate, descriptive titles and keywords. Don't worry about imitators. 

[daniellei4510]

Totally agree. I will often copy my own prompts (not titles) that generated successful images previously and the results rarely come even close to the original iterations. So I'm not reallly sure what's going on in these examples being as close as they are in resemblance. AI results are just too random even with the same prompt, much less using a title describing the image after the face. 

[Abambo]

So I'm not reallly sure what's going on in these examples being as close as they are in resemblance. AI results are just too random even with the same prompt, much less using a title describing the image after the face. 

By @daniellei4510

You use a reference image.

[daniellei4510]

That would increase the odds, but that closely? Possibly. I've played with referencing my own images from my photography days and the results are are at least closer to the originals while still far removed. A photo of my 8 year old daughter, for example, yielded adult women in various poses, while the colors and overall style were retained.

[Abambo]

As long as you do not know the AI engine they use… it will be difficult to find definitive answers.

[daniellei4510]

There's that, plus the examples in question are all very geometric, which would be easier to duplicate with a reference images.

[daniellei4510]

I guess the bottom line is this: obtaining and uploading a reference image and ripping off someone's titles is certainly more work than just coming up with one's own prompts. There are even sites on Facebook where sharing prompts is welcomed and encouraged, at least for instructive purposes, and from which variations can be experimented with. Why someone would even take the time to do this is mind-boggling. Probably watched too many get rich quick videos on the Tube.

[Abambo]

Adobe do not care about this.

By @TatyM

So stop contributing.

[TatyM]

Thak you for your advice

[Abambo]

You're welcome.