Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
×
AI Businesses

Getty Images CEO Says Content-Scraping AI Groups Use 'Pure Theft' For Profit (fortune.com) 64

Getty Images CEO has criticized AI companies' stance on copyright, particularly pushing back against claims that all web content is fair use for AI training. The statement comes amid Getty's ongoing litigation against Stability AI for allegedly using millions of Getty-owned images without permission to train its Stable Diffusion model, launched in August 2022.

Acknowledging AI's potential benefits in areas like healthcare and climate change, Getty's chief executive argued against the industry's "all-or-nothing" approach to copyright. He specifically challenged Microsoft AI CEO Mustafa Suleyman's assertion that web content has been "freeware" since the 1990s. The Getty chief advocated for applying fair use principles case-by-case, distinguishing between AI models for scientific advancement and commercial content generation. He also drew parallels to music streaming's evolution from Napster to licensed platforms like Spotify, suggesting AI companies could develop similar permission-based models.

He adds: As litigation slowly advances, AI companies advance an argument that there will be no AI absent the ability to freely scrape content for training, resulting in our inability to leverage the promise of AI to solve cancer, mitigate global climate change, and eradicate global hunger. Note that the companies investing in and building AI spend billions of dollars on talent, GPUs, and the required power to train and run these models -- but remarkably claim compensation for content owners is an unsurmountable challenge.

My focus is to achieve a world where creativity is celebrated and rewarded AND a world that is without cancer, climate change, and global hunger. I want the cake and to eat it. I suspect most of us want the same.

Getty Images CEO Says Content-Scraping AI Groups Use 'Pure Theft' For Profit

Comments Filter:
  • Let me try: By reading this comment you agree to pay me $400. If you don't agree with this, stop reading now.

    If you're still reading I'll expect those smackaroos. Failure to pay me is pure theft. I'll prove it by noting anything similar you ever say to the following random word list and suing you if I find them.
    finger
    link
    husband
    brilliance
    chew
    surgeon
    gas
    contemporary
    recovery
    fibre
    Because the transformative work going on in your brain is pretty close to what happens with an LLM and because you're capable of r
    • No problem . I read the list. I will wire $400.00 to you in ShitCoin (400Trillion Shitcoin) to you. However, Our terms of service require you to give your wallet address, password, and the 12 magic words. Nice grift, thought I'd get in on it too. I love how term of service can demand anything these days. Your bank account numbers. Your Health status. Your First born. This IS fun. :-)
    • Let me try: By reading this comment you agree to pay me $400. If you don't agree with this, stop reading now. If you're still reading I'll expect those smackaroos. Failure to pay me is pure theft. I'll prove it by noting anything similar you ever say to the following random word list and suing you if I find them. finger link husband brilliance chew surgeon gas contemporary recovery fibre Because the transformative work going on in your brain is pretty close to what happens with an LLM and because you're capable of remembering the list.

      I'm unable to produce a response.

    • A human reading anything is not considered making a copy under copyright law. If said human were to write it down verbatim, then and only then is it copyright infringement.
  • by Forty Two Tenfold ( 1134125 ) on Monday December 02, 2024 @02:19PM (#64985697)
    OK, so I can use Microsoft content for free as well? Or just freely watch the ads they serve me on MSN?
    • I found a bunch of Microsoft software on the internet, must all be freeware.

    • Re:Freeware. (Score:4, Informative)

      by fafalone ( 633739 ) on Monday December 02, 2024 @04:40PM (#64986083)
      Yes, you can go to microsoft.com, and read thousands of pages for free. You can even learn from them and remember their content without paying, and even paraphrase what you've learned for others, even quoting small portions! Only when you distribute large portions verbatim do they expect payment. When we can extract data from the brain I look forward to all the copyright lawsuits for the training material you scraped with your eyes, since there will now be some mechanism by which a 3rd party can bypass your lack of intent to distribute large verbatim content.
      • I believe the Microsoft Dude said that Microsoft software is free to grab all over the internet, not just from the obvious web address, and anytime since the 1990s. I believe he also said it's ok if any non-Microsoft person secretly uploads any Microsoft software onto a private server, so that it can be downloaded for free. Because as long as there's no impediment to downloading, that means Microsoft has agreed to let everyone have it for free forever.

        Just don't download MS-DOS please, because that's sof

  • but more importantly, the nightmarish AI reels you see on instagram

  • ...about robots creating crap
    Clip art and stock photos never made anything better

  • Stealing (Score:4, Interesting)

    by Anonymous Coward on Monday December 02, 2024 @02:30PM (#64985723)

    The same Getty Images that demands payment for public domain images and images they don't own

  • True (Score:3, Insightful)

    by GameboyRMH ( 1153867 ) <gameboyrmh@Nospam.gmail.com> on Monday December 02, 2024 @02:31PM (#64985725) Journal

    The AI industry is industrial-scale piracy for profit with just enough obfuscation to allow the companies to become too big to jail at launch. I doubt we'll see any lawmakers calling for these companies' ISPs to be held accountable for their behavior though.

    • The AI industry is industrial-scale piracy for profit with just enough obfuscation to allow the companies to become too big to jail at launch. I doubt we'll see any lawmakers calling for these companies' ISPs to be held accountable for their behavior though.

      The law is set up to protect the owner class. AI in its current iteration is a smokescreen for the owner class to take complete ownership of the web. Forget about open communication. It'll be flooded with AI derived crap until there's nothing left but rot and lawsuits for daring to use AI generated content without paying the owner class for the right. Progress!

      • This is wrong, there are many free models you can use. But even if you use closed models, the benefit sits with you as the user. You set the task, you give the inputs, you give the feedback, and you apply the ideas in your own life. OpenAI makes cents / million tokens, running at a loss. The beneficiaries of AI are whoever use AI because they get the outcomes.
        • This is wrong, there are many free models you can use. But even if you use closed models, the benefit sits with you as the user. You set the task, you give the inputs, you give the feedback, and you apply the ideas in your own life. OpenAI makes cents / million tokens, running at a loss. The beneficiaries of AI are whoever use AI because they get the outcomes.

          And you miss the forest for the trees. The concept is that you, through use, are training this thing to do whatever task it is you are trying to accomplish, so that someday, some "higher up" can accomplish that task without you. Benefit now? Questionable. Benefit later? Someone else's, if the fantasy fetishists pushing this computer god concept are right.

    • Absurd. It is far more clear that the current incarnation of copyright law is garbage in outrageous violation of Article 1 Section 8 that claims to authorise it. Disney, specifically Jack Valente, spent his life pushing a wildly perverse interpretation of property rights that he was able to say loud enough and repeat often enough people actually believe it uncritically. The law hindering AI technology is just the next iteration of the law harming progress.
      • I agree that copyright law is garbage that's harming progress, but I'd prefer if it was applied consistently.

        • I agree that copyright law is garbage that's harming progress, but I'd prefer if it was applied consistently.

          I look for the next update to copyright law to take some of the ideas free-thinkers espouse "limit copyright to seven years" to small producers, and corporation owned copyright to be forever plus infinity, to justify the Disney types stealing creative output from actual talent, then "owning" it outright forever. We've hit a point where the corporations control not only the government, but also the public narrative for the most part. People will cheer when the last of their rights are trampled into oblivion.

    • Piracy is redistributing copies of the original. AI Training is not.
    • by Tora ( 65882 )

      You obviously don't understand copyright law, and the nature of derivative vs transformative works.

  • The same people who protest DRM should not protest scraping, as ripping a web site is the same as ripping a CD in terms of data. The only difference is that the amount of data a single human can read in their lifetime versus an all powerful AI. As bandwidth gets wider (Datacenters are in the terabit range) and data storage gets cheaper (A petabyte will be available for less than $10K soon) AI will be able to consume all possible human output. Like it or not it will overwhelm the economy. You can either smas
    • Then AI will be trained on AI
      Will that end up like mad cow disease? Infection spreading through consumption.

    • by SirSlud ( 67381 )

      "The only difference" is the point. A lot of law is predicated on intent and consequence. What's the difference between stealing 5 dollars and 5 million dollars? Well, nothing, it's both stealing money .. and yet, we realize they're two different acts because of the difference in scale.

      • Now that you mentioned intent, why is everyone forgetting the new intent contributed by the user by prompting? The models don't act of their own self intention, they are called to solve specific tasks which are dependent on their users. The task is not copying, and if that was the task, then it's simpler to just .. copy. Don't need a LLM for copyright infringement, it's actually a bad tool for infringing. The user intent is what deviates the model from simple replication of its training data.
    • Yes, but it is learning a model of language, not storing those sources exactly. The model itself is 1000x smaller than its training set, it can't possibly store it verbatim. It is compressing that data into reusable abstractions and composable skills. On top of that, the model gets human prompts to guide it, so new information blends with old information at inference time. It is not simply regurgitating, but adapting what it learned to the context and new data it receives.
  • .. law works.

    By his logic, if I find a Windows 11 ISO and Activation Key somewhere online, it is "freeware" and I can do anything i want with it.

    Of course, that is pure baloney, because Microsoft owns the copyright regardless of how I come across it, and Microsoft dictates the terms around how I can license that content.

    There is zero difference whatsoever in copyright law between

    - A piece of software
    - An image
    - A book
    - A blog post

    They are all creative works, and all afforded identical status in copyright law.

    • So Microsoft can come knocking at my door any time and demand payment for all the income Ive derived from the tools and documentation they freely posted on their website I then used to make money by applying the tools and knowledge? Or only if Ive saved something to reuse after they delete it? Because I've crossed some barrier where I just made too much money from what they gave away for free in ways that should have clearly fallen under fair use until they saw dollar signs? I just don't see a difference be
      • They say models can ingest much more data than a human, but they are wrong. A human with a search engine has the whole internet to pick from without having to read all of it.
      • by brunes69 ( 86786 )

        You would feel differently if you found out that Microsoft, OpenAI, and Google were making billions of dollars off of something you created, and *did not post for free*, but licensed out to individuals.

        That's what these photographers (who Getty represents) are suing for. They never gave consent to any of this.

    • by SvnLyrBrto ( 62138 ) on Monday December 02, 2024 @06:41PM (#64986427)

      But he does obviously remember the old school internet and www culture from the 1990s. It didn't start out this corporatized and sanitized and controlled and DRM'd thing it's been allowed to become. The "social contract" on the web was originally that when you put anything up on port 80, you were channeling William Mulholland and telling the world: "There it is. Take it." You were expected to give credit and put your own stuff up on the web for others to use in return, of course. But no one kicked in Craig Peters' or Lars Ulrich's or Hillary Rosen's door and took their stuff. They and their ilk barged into other people's pre-existing potluck and put their dicks in the mashed potatoes.

      • by brunes69 ( 86786 )

        Good thing "social contracts" do not have the force of law.

        Also, once again - if I find a copy of Windows 11 and it's license key on "port 80", that apparently means it is free game by your logic. I am sure Microsoft would agree with that as well?

    • No, if you find a bunch of code online and learn how to code from it, you then can get a living as a developer and don't have to pay the authors of those codes anything. There is a process of abstracting away ideas and repurposing them in this loop. If the job of a developer was just to copy paste, we wouldn't need to pay them so much. And if LLMs were only good at copying existing ideas, then Google Search has been helping us do the same for decades.
      • by brunes69 ( 86786 )

        Ah, the old doctrine of "I found it, therefore, I own it"

        Good thing that isn't how copyright law works, at all.

  • These days Getty's just a clearinghouse for meaningless stock imagery, the sorts that will very soon be replaced by AI-generated schlock.

    You know, all those smiling health-care professionals, and that IT dept. that looks like it's around 20 years out of date.. all those placid, pleasant images you see in advertisements are 99.9999% stock photos bought (or taken) from places like Getty.

    All of that will be AI inside 3 years from now.

    Getty's only value is in their historical photographs.

    • by brunes69 ( 86786 )

      The only way that AI knows how to make any of those photographs, is because it was trained on Getty, and therefore they should be compensated.

      That is the whole legal argument here.

      • No, each and every photog who took stock pix and sent 'em to getty only to get a fraction of a penny for every time such a picture is used are the ones who should be compensated.

        Getty, the in-betweener, can go pound sand. They've already made their fortune on the backs of the great bulk of never-was photographers who contributed to their trove.

        • by brunes69 ( 86786 )

          Er....

          How do you think Getty pays their 45% royalties to the photographers, if they don't get paid by OpenAI and Stability?

          That is *EXACTLY THE POINT OF THE LAWSUIT, GENIUS*

          • Let them be paid for replicating their images, not for training on their data and generating different images. They don't own abstract topics like "smiling IT professional answering phone" or whatever generic shit they have. They own images, not abstractions.
      • How can this legal argument hold? Are we putting royalties on deriving benefits from copyrighted materials when no protected expression is being replicated? That's new. Copyright is about protecting expression against exact replication, not to extract revenue from all possible benefits. If I owe my life to reading a book about survival in the wild, should I become the salve of who wrote the book?
    • No, generative AI is good at taking your one photo and making it better, giving you variations, etc. So you can start from your real product and generate extra/refined images from it.
  • by bradley13 ( 1118935 ) on Monday December 02, 2024 @02:59PM (#64985795) Homepage
    Scummy company, scams their own customers. Get fscked.
  • #irony (Score:5, Interesting)

    by Travelsonic ( 870859 ) on Monday December 02, 2024 @03:26PM (#64985853) Journal
    Coming from the company that has gotten in trouble dor nicking others' photos before, tries to license out public domain images all the time (loads of PD images on their site not just with their watermarks, but with "licensing options), who IIRC even owned the subsidiary at the center of controversy over trying to claim ownership of an image of the corona virus.

    Yeah no, fuck off Getty ya two-headed cunt.
  • by Fly Swatter ( 30498 ) on Monday December 02, 2024 @04:10PM (#64985985) Homepage
    The internet has changed 'fair use' and 'copyright', deal with it.
    • Well said, it all started 25 years ago. Web search looks like prompting if you slant. You put in some text and get out some text. Image search - well with billions of images it's almost like prompting a diffusion model, whatever you search for rule 34 allows to exist. Chatting in forums? it's like chatting with LLMs too. So it seems search and social networks have been doing what LLMs are doing but with a network of people instead of a model. A manual AGI or HGI (human GI).
  • Was about time. It strayed way too far from the original purpose
  • Stop trying to take those tasks away from humans. Where's my AI that does my taxes? Or finds me a new job? Or does the dishes? That's what AI should be doing: Freeing humans from drudgery.
    But, of course, those are things that people monetize, so we can't have AI do that...

    • AI will be doing your new job, not finding it for you.
      • Why? AI can invent work as well, not just solve it. When you have a new capability you create an ecosystem of work around it, and humans are strictly necessary for now. We don't know how to make AI useful without human in the loop. It's still a tool not an autonomous agent.
        • In various expertise domains and skill areas, AI and automated systems / robotics will cross over a better and/or more cost-effective than humans threshold. This will happen domain by domain, job by job, skill by skill. And the pattern may be that for some jobs and skills, the AI is only enhancing the quality and productivity per person, rather than completely replacing the human. Nevertheless the human component of the work is thereby economically devalued i.e. human + AI produces what human alone produced

Every little picofarad has a nanohenry all its own. -- Don Vonada

Working...