Anti AI scraping options

[sammey]

Hi Guys, 

Is anyone aware of the development in Lightroom or Photoshop to introduce anti scraping technologies into metadata? 

[jane-e]

Is anyone aware of the development in Lightroom or Photoshop

By @sammey

If it is in development, then anyone with knowledge would be under NDA and could not say.

Jane

Forum volunteer

[sammey]

Thanks Jane, well I certainly hope they are - it's very important to me as a creator that my work goes out with my IP protected. I'm surprised to be honest that something hasn't already been deved and integrated.

[jane-e]

You're welcome @sammey. Your post is in Ideas where the developers will see it and like-minded users can upvote it and comment.

Jane

[Stephen Marsh]

I have some ExifTool commands and Bridge and Photoshop scripts... That being said it's next to useless, it's an honour system and the key players have no honour for I.P. or existing copyright laws, let alone new initiatives.

I'll dig up the links...

EDIT:

https://community.adobe.com/t5/photoshop-ecosystem-ideas/add-support-for-iptc-plus-coalition-generat...

[ExUSA]

Yes there are scripts to add this but right now its a legal and practical grey area.

[M-H]

I highly doubt there would be anything like it anytime soon, even just because of read & write properties of metadata. If you can read it, you can also write whatever you like in there. If you are that worried about your work, you should look into get it protected by actual law/contracts/usage argreements and not just by a tick box in file settings

[sammey]

Thanks so much @Stephen Marsh 

[sammey]

Thanks for your reply  @M-H but contraccts with clients won't stop AI training programs from scraping my work. Certain website platforms (like Squarespace) have an option to tell scrapers the website is off limits. ' I know Cloudflare has deved several features to prevent (what I would term) malicious scraping - ie copyright theft. I'm looking for a process like this that starts at source. A digtal watermark or a money dye pack (if you like) even at pixel level that protects your work. I hope that Adobe is working on a solution as I'm sure there will be a huge demand for it with current legistlation being passed by various Governments in the UK, US and Australia granting AI companies free reign to help themselves to creators IP.  The tech is beyond me but I'm hoping someone out there is working on it if not Adobe. 

[M-H]

@sammey I get what you are saying but I just cannot even remotely imagine how something like this would be executed. Even if you had solid proof (I also cannot imagine how would you go abtaining such a proof) that company XYZ has used your website to train their model, are you actually going to take them to court? What would be the claim? You loosing some income by them doing that? Can you imagine what sort of legal teams they have at their disposal? Also bigger AI companies are already stating their image models were trained on licensed or public domain image sets (Adobe on Adobe Stock, Bria on Getty Images, Alamy etc) so its legally cleared.

[jane-e]

I hope that Adobe is working on a solution

By @sammey

Your post is in Ideas and will be seen by the developers. We don't know what they are working on until they announce it. I've upvoted your initial post.

Jane

[sammey]

It's been difinitively proven that generative AI companies have scraped copyrightted material to train their software. It's yet to be definittely proven that this is legal despite the AI companies using transormative data as an excuse - as per the mutlipe legal challenges currently going on. It sounds like you are OK with that (and I'm fine with you being OK with that) - but I put an enourmous amoutn of blood, swear and tears into my work - and I am not OK with it. Just because a big AI company says something, it does not neceassirly make it so. My original post - is to do with creators being given the option to tag thier work with an effective means of letting AI scrapers know - that the creator of that work doesn't want thier work to be stolen for proft. I'm not stupid - I understand it may be next to impossile to enforce - but that isn't a reason to do nothing.  

[c.pfaffenbichler]

»That being said it's next to useless, it's an honour system and the key players have no honour for I.P. or existing copyright laws, let alone new initiatives.«

Harsh, but hard to argue … 

[sammey]

@c.pfaffenbichler   absolutely with you on that. But you never know what the future holds I guess.. If you had a system that could track your work - however technically improbable that may sound - and a way to audit the training programs in the future (if there polictal will changes), maybe, just maybe there's could be a mass tort action down the pipe. Who know's maybe AI will come up with some software that makes the impossible possible - wouldn't that be an irony? I find it interesting that those who have blind faith in tech - are the same people saying the tech doesn't exist to handle complicated concepts like some sort of money bag dye on a pixel level, a digital,  indellible mark of ownershiip. I think there's a strong argument to stamping your work now - even if check/ tracking systems aren't currently in place - they maybe in the future. Either way I'm all for it as an option at least. Both sides should be served, those that are pro AI and those that aren't. All I'm asking for is a slide button... 

[M-H]

I totally get the need for a button/function like that and I'd probably use it myself, but I would never expect that to be 100% solid protection. Photoshop ages ago used to come with Digimarc for digital watermarking and im guessing if it was any good/usefull back then people would be still using it now...

But you never know what the future holds I guess..

True that, but why assuming the worst is going to happen?

[sammey]

I think it's healthy to assume the worse, try and have a contigency in place, then be pleasantly surprised if it doesn't come to pass. My perssimistm is largely based on the track record the social media giants have in terms of money over ethics, hiding behind free speach legistlation to spread disinformation and social discord then not paying the butchers bill -  the list seems pretty endless. I can't see AI companies behaving any better - especially when the starting point is copyright theft. I have nothing against ai and ai creative tools that have been trained ethically and responsibly, where permissions have been given and / or compensation awarded. Barnoness Kidron did an amazing job in the House of Lords fiighting for transparency over AI learning sources bcoming legistlation but the Government shut her down. 

As a freelance photographer operating in and very tough commercial climate, I am completely opposed to AI companies helping themselves to my work. It's like McDonalds making and selling burgers from rustled cattle, straight up illegal and unethical. I've shot personal projects that have cost me money, time and huge effort to bring into the world - and it makes my blood boil to think that this copyrighted material is considered fair game by tech oligarchies who's only motive is profit. So even if it's kinda pointless, I'd still hit that slider... 

[davescm · Accepted solution]

@sammey You might want to look at these.

Some info on the Content Authenticity Initiative : https://contentauthenticity.org/

Info on the Content Credentials option that can be turned on in Preferences > History and Content Credentials in the current version of Photoshop (v26.9).
https://helpx.adobe.com/uk/photoshop/using/content-credentials.html

Dave

[sammey]

Thanks @davescm I appreciate it I'll take a look.