New Adobe Terms

[Nimi]

I won't try to explain in details the terms just issued (then explained 24 houts later) by Adobe; my interpretation is that they are giving themselves the right to access and scan for AI everything that's on their clouds, including what we upload to Lightroom, Photoshop, Illustrator, etc. without our consent or ability to opt out. There has been ton of reporting on the topic in the last 48 hours, Google it.

Practically, I've been instructed by a couple of my agencies to stop using AI for retouching (inc. Generative Fill and that magic erase wand) AND start shipping some assets to them via disk, not (any) cloud. Reason for the former is that there is an assumption that using these tools is essentially entrapment (my word) and those photos are now fair game for AI scanning. Reason for the latter is that they are no longer confident that photos of products that haven't been released yet wouldn't end up somewhere prior to official release.

Over-reaction, maybe. I used to perceive Adobe as "good guys," but that perception is quickly eroding.

 

[TurtleCat]

I was a bit surprised to see them do something this bold without an opt-out. I don’t think pressure will change their minds, either, as it didn’t with subscriptions. They will probably assume most people won’t bother to read the terms, understand them, or be motivated enough to find alternatives. It’s the Wild West right now with machine learning and related toolsets. I suspect regulation will eventually catch up but the horse has left the barn and is in another country already. I understand Facebook is doing the same thing but they did make an opt out that’s very difficult to find and configure. I suspect that this may help vendors like Topaz who don’t upload to the cloud for their generative erase and enhancements to some extent.

 

[pomkiwi]

Adobe did this last year for images uploaded to their stock platform for sale (I think that it had been in the terms for some while before last year). To be fair to Adobe they did make a payment to all contributors to their stock platform to reflect this usage, which I felt was fair particularly as I'm not sure other platforms did the same. There was a vague suggestion taht further payments might be made in the future but I'm not sure whether that is likely.

 

[Hooligans Imagery]

Adobe has already clarified it. As Taylor Swift would say "you need to calm down." "You" meaning their users.

Our commitments to our customers have not changed.

• Adobe does not train Firefly Gen AI models on customer content. Firefly generative AI models are trained on a dataset of licensed content, such as Adobe Stock, and public domain content where copyright has expired. Read more here: https://helpx.adobe.com/firefly/faq.html#training-data
• Adobe will never assume ownership of a customer's work. Adobe hosts content to enable customers to use our applications and services. Customers own their content and Adobe does not assume any ownership of customer work.

 

[Nimi]

Adobe has already clarified it. As Taylor Swift would say "you need to calm down." "You" meaning their users.

Our commitments to our customers have not changed.

• Adobe does not train Firefly Gen AI models on customer content. Firefly generative AI models are trained on a dataset of licensed content, such as Adobe Stock, and public domain content where copyright has expired. Read more here: https://helpx.adobe.com/firefly/faq.html#training-data
• Adobe will never assume ownership of a customer's work. Adobe hosts content to enable customers to use our applications and services. Customers own their content and Adobe does not assume any ownership of customer work.

Far more complicated than this and Adobe has not walked back the following. Academic for me because what I do is dictated by my clients, many of whom are not in a trusting mood.

 

[TurtleCat]

Another good summary of what has and hasn’t been included in this saga: https://appleinsider.com/articles/2...rms-but-the-damage-may-have-already-been-done

 

[Hooligans Imagery]

Far more complicated than this and Adobe has not walked back the following. Academic for me because what I do is dictated by my clients, many of whom are not in a trusting mood.

They did clarify that; open the link, please. It is for content review for inappropriate material. I trust that they are not fools and understand the court of public opinion as well as everything else.

 

[Nimi]

They did clarify that; open the link, please. It is for content review for inappropriate material. I trust that they are not fools and understand the court of public opinion as well as everything else.

Lets quote it verbatim: “To be clear, Adobe requires a limited license to access content solely for the purpose of operating or improving the services and software and to enforce our terms and comply with law, such as to protect against abusive content." The first part, "for the purpose of operating or improving the services and software" includes Firefly and "operating or improving" would include scanning and feeding models. Later they say that they don't own users' content. Duh. Scanning does not require ownership. Under current copyright law it's 100% legal to scan copyrighted content and use it in LLMs.

I think they are simply admitting what the rest of the AI industry has been doing: "if it's not nailed down, it's ours for the taking. If we can pry it loose, it's not nailed down."

 

[Hooligans Imagery]

We can leave it to the lawyers and public to decide. It's a big enough user group that this will get sorted out.

 

[Nimi]

Here is my take:

1. If you post online, your content has been scanned by LL models. Perfectly legal and there isn't a way to stop it.

2. Your scanned content is being used to create new content.

3. If you use AI tools (eg Adobe Generative Fill) that require an upload into the company's servers (eg Adobe Generative Fill) there is a very good chance those images are also being used for models.

This may or may not matter to you.

 

[ButlerKid]

Thank you for posting this. I do not store any of my images on the cloud.....except for my SmugMug account.

However, I have recently been using PS fill and remove tools on occasion. Not being a professional or someone trying to earn money from my images, I will have to decide whether or not I want to change my current workflow.

 

[Dave Stargazer]

Here is my take:

1. If you post online, your content has been scanned by LL models. Perfectly legal and there isn't a way to stop it.

2. Your scanned content is being used to create new content.

3. If you use AI tools (eg Adobe Generative Fill) that require an upload into the company's servers (eg Adobe Generative Fill) there is a very good chance those images are also being used for models.

This may or may not matter to you.

Thanks for posting, Nimi - this too has been my take for a while, regardless of what every AI-adjacent tech company says publicly - if you post it, it will get scanned whether you give permission or not. It's one reason I include a (very subtle) watermark on everything now even though I know it can be removed, or ignored - it was still part of the original scan/download of my work, and that very possibly that fact will remain in their intake chain.

I have a feeling Adobe is going to come back with more "clarification" to try and backtrack, and comfort their user base. But uploading to their cloud pretty clearly makes your work fair game. I slot Adobe in the same category as any large corporation - i.e. assume the negatives unless proven otherwise - and that's been true for Adobe since their inception; but they have been very good at knowing their customers, and generally produce good/great products.

Another defensive thing to do is never post full resolution (I don't except for very rare tight crops to demonstrate resolving power) - the small the better.

Some of my best work I hold in reserve, and will never get posted online without strong encryption. That includes all aspects of the works I produce in the various domains I work in.

Cheers!

 

[Jerry Watson]

Makes me wonder if LR cloud users will switch to LrC.

 

[John Navitsky]

one thing to consider is i suspect they may need to upload, process and save your image to provide some of the Content Credentials functions, esp the part that allows id of images that have had their CC data stripped as well as the before/after functions

 

[Hooligans Imagery]

one thing to consider is i suspect they may need to upload, process and save your image to provide Content Credentials functions

John yes another good example. One really has to read their entire response that I linked above for perspective. Most of the people responding to the terms aren't trained in legal matters. From my career years of tax controversy, one can't just pick out part of a sentence. One has to look at the entire context. People would complain the other way - let's talk about the example on the page.

but we may use available technologies, vendors, or processes, including manual review, to screen for certain types of illegal content (for example, child sexual abuse material) or other abusive content

If they didn't have the proper access, then those posting illegal content could be let off.

It's complicated and requires understanding the entire agreements and context. I personally don't put Adobe in the context of those giving us services so they can sell advertising. I think this will get sorted out appropriately. The content owners like the movie producers and Adobe will talk I am sure and adjustments will be made.

 

[Cropfield]

Here is what Tony Northrup has to say about Adobe's updated terms (June 7th):

 

[bleirer]

Yoda says, The click bait is strong in this one.

 

[Nimi]

John yes another good example. One really has to read their entire response that I linked above for perspective. Most of the people responding to the terms aren't trained in legal matters. From my career years of tax controversy, one can't just pick out part of a sentence. One has to look at the entire context. People would complain the other way - let's talk about the example on the page.

If they didn't have the proper access, then those posting illegal content could be let off.

It's complicated and requires understanding the entire agreements and context. I personally don't put Adobe in the context of those giving us services so they can sell advertising. I think this will get sorted out appropriately. The content owners like the movie producers and Adobe will talk I am sure and adjustments will be made.

While there are legit reasons for them to access our files, wherever they are, they have clearly worded it to allow them to use our images to train Firefly. It may not matter to many; my agencies care and are taking steps to minimize this (legal) incursion.

 

[John Navitsky]

While there are legit reasons for them to access our files, wherever they are, they have clearly worded it to allow them to use our images to train Firefly.

and yet, they flat out said they won’t

while all of us, including your clients should do what they think is best, i will point out that adobe seems to be one of the most careful in terms of respecting ip in their ai

and further, they are driving the Content Credentials effort which is vastly important for protecting content creators and for truth in general

 

[John Woodworth]

Here is what Tony Northrup has to say about Adobe's updated terms (June 7th):

This video seems to be a bit hysterical. However, it does raise, IMO, realistic issues arising from giving human access to so much personal data.

No matter how high-minded the motivation, someone with "privileged access" will ultimately rationalize the most bizarre and unethical abuse. From the petty--like the Sys Admin who uses special privileges to avenge a personal slight--to the consequential--like Edward Snowden who revealed government excesses (for better or worse)--it seems that society has yet to come to grips with the tradeoff's associated with aggregating large amounts of personal data.

So maybe a little hysteria is justified.

 

[Nimi]

and yet, they flat out said they won’t

while all of us, including your clients should do what they think is best, i will point out that adobe seems to be one of the most careful in terms of respecting ip in their ai

and further, they are driving the Content Credentials effort which is vastly important for protecting content creators and for truth in general

Where does it say they won't?

 

[Not A Speck Of Cereal]

Makes me wonder if LR cloud users will switch to LrC.

For the last half-to-full year, there have been 'agents' on social media accounts asking users to justify why they haven't yet moved from LrC to LR Cloud. (Typical arguments between feature sets ensued, not germane to this post.)

The first agents openly admitted they represented Adobe.

After that first wave, it was people claiming to be independent or not openly mentioning they were representing any entity, but still seemed to use the same language and ask the same questions.

At the time, I thought "As Adobe is likely behind all of this, it's an attempt to sell more cloud storage" and ignored it, except for the annoying thoughts that LrC may one day go away (I don't store my content in the cloud, so if they drop LrC, I will drop Adobe ... probably for C1).

But NOW I'm starting to believe that these were possibly attempts to get more content in the cloud (from both recreational and commercial users) for reasons related to this thread. Suffice it to say that I'm concerned. And I will mostly certainly not "calm down", because I've spent decades in the corporate world and know how they think.

Chris

 

[John Woodworth]

Where does it say they won't?

Speaking from first-hand observation, if it doesn't say they won't, they will. And even if it does say, they still will.

 

[John Navitsky]

not that anyone cares, but i’ll never click on a Tony Northrup video after he accused Steve McCurry of misconduct over the Afghan Girl photo without the evidence such accusations require. it’s hard to see now what exactly transpired there since he was allowed to drop in revised videos so what we see now is a revisionist history, but some of us remember what was very poor behavior and not worthy of rewarding with clicks or other support

ymmv

 

[John Woodworth]

For the last half-to-full year, there have been 'agents' on social media accounts asking users to justify why they haven't yet moved from LrC to LR Cloud. (Typical arguments between feature sets ensued, not germane to this post.)

The first agents openly admitted they represented Adobe.

After that first wave, it was people claiming to be independent or not openly mentioning they were representing any entity, but still seemed to use the same language and ask the same questions.

At the time, I thought "As Adobe is likely behind all of this, it's an attempt to sell more cloud storage" and ignored it, except for the annoying thoughts that LrC may one day go away (I don't store my content in the cloud, so if they drop LrC, I will drop Adobe ... probably for C1).

But NOW I'm starting to believe that these were possibly attempts to get more content in the cloud (from both recreational and commercial users) for reasons related to this thread. Suffice it to say that I'm concerned. And I will mostly certainly not "calm down", because I've spent decades in the corporate world and know how they think.

Chris

I have observed the same and have the same concerns.