An image of an archeologist adventurer who wears a hat and uses a bullwhip
(theaiunderwriter.substack.com)1436 points by participant3 2 days ago
1436 points by participant3 2 days ago
> The ONLY reason to have any law prohibiting unlicensed copying of intangible property is to incentivize the creation of intangible property.
That was the original purpose. It has since been coopted by people and corporations whose incentives are to make as much money as possible by monopolizing valuable intangible "property" for as long as they can.
And the chief strategic move these people have made is to convince the average person that ideas are in fact property. That the first person to think something and write it down rightfully "owns" that thought, and that others who express it or share it are not merely infringing copyright, they are "stealing."
This plan has largely worked, and now the average person speaks and thinks in these terms, and feels it in their bones.
>the average person speaks and thinks in these terms,
(Trademarks aside) Even more surprising to me is how everyone seems concerned about the studios making enough money?! As if they should make any money at all. As if it is up to us to create a profitable game for them.
If they all go bankrupt today I won't lose any sleep over it.
People also try to make a living selling bananas and apples. Should we create an elaborate scheme for them to make sure they survive? Their product is actually important to have. Why can't they own the exclusive right to sell bananas similarly? If anyone can just sell apples it would hurt their profit.
It is long ago but that is how things use to work. We do still have taxi medallions in some places and all kinds of legalized monopolies like it.
Perhaps there is some sector where it makes sense but I can't think of it.
If you want to make a movie you can just do a crowd funder like Robbert space industry.
> Even more surprising to me is how everyone seems concerned about the studios making enough money?! As if they should make any money at all. As if it is up to us to create a profitable game for them.
Do you want more games (movies, books...)? Then you want studios to make money in that type of game. Because and if they make money they have incentive to do so. Now if you are happy with the number and quality of free games a few hard core people who will do it even if they make nothing then you don't care. However games generally take a lot of effort to create and so by paying people to make them we can ensure people who want to actually have the time - as opposed want to but instead have to spend hours in a field farming for their food.
Now it is true that games often do look alike and many are not worth making and such. However if you want more you need to ensure they make money so it is worth investing.
We can debate how much they should make and how long copyright should be for. However you want them to make money so they make more.
It's been a US-led project for the benefit of American corporations.
If I was running the trade emergency room in any European state right now, I'd have "stop enforcing US copyright" up there next to "reciprocal tarrifs".
We were close to your viewpoint being the popular one, but sadly many (most?) independent content creators are so overtaken by fear of AI that they've done a 180. The same people who learned by tracing references to sell fanart of a copyrighted franchise (not complaining, I spend thousands on such things) accuse AI of stealing when it glances at their own work. We're entering a new golden age of creative opportunity and they respond by switching sides to the philosophy of intellectual property championed by Disney and Oracle (except for those companies' ironic use of AI themselves..).
We would prefer a world where we can use the skills we have spent a lifetime honing without having to compete with some asshole taking everything we’ve shared and stuffing it into a machine that spits out soulless clones of our work without any acknowledgment of our existence.
> we were close
Maybe. In my microcosm even before big AI, 100% of my tech acquaintances were against IP laws, 0% of my art acquaintances were, and authors I know had varied opinions based on their other backgrounds.
Artists do seem to have had a mindset shift. Previously they supported IP protection because it was "right" (or they'd at least concede that in practice it's not helping them personally), but with the AI boom most of them are pro-IP laws because of more visceral livelihood fears.
There's also a moral issue at play: To safeguard the interests of a few publishers (sometimes the creators, but they can easily end up with a shitty deal) you remove freedoms to the entire population to copy the same idea.
You need a central structure funded by everyone's taxes which enforce a contract almost nobody of the infringers has signed.
That's appaling, I hope with this AI wave we'll get rid of copyright all together.
Trademark isn't copyright, those are two different things. Trademarks can be renewed roughly every 10 years [1] until the end of time and are about protecting a brand. Now copyright law lasts for "author plus 70 years. For anonymous works, pseudonymous works, or works made for hire, the copyright term is 95 years from the year of first publication or 120 years from creation, whichever comes first." [2]
Is copyright too long? Yes. Is it only that long to protect large media companies? Yes. But I would argue that AI companies are pushing the limits of fair use if not violating fair use, which is used as a affirmative defense by the way meaning that AI companies have to go to court to argue what they are doing is okay. They don't just get to wave their hands and say everything is okay because what we're doing is fair use and we get to scrape the world's entire creative output for our own profit.
[1] https://www.uspto.gov/learning-and-resources/trademark-faqs#...
[2] https://www.copyright.gov/history/copyright-exhibit/lifecycl...
Trademark isn't the same as Registered Trademark either, while we're at it
> There are approximately zero people who decide they'll create something if they're protected for 95 years after their death but won't if it's 94 years.
I’m sure you’re right for individual authors who are driven by a creative spark, but for, say, movies made by large studios, the length of copyright is directly tied to the value of the movie as an asset.
If that asset generates revenue for 120 years, then it’s slightly more valuable than an asset that generates revenue for 119 years, and considerably more valuable than an asset that generates revenue for 20 years.
The value of the asset is in turn directly linked to how much the studio is willing to pay for that asset. They will invest more money in a film they can milk for 120 years than one that goes public domain after 20.
Would studios be willing to invest $200m+ in movie projects if their revenue was curtailed by a shorter copyright term? I don’t know. Probably yes, if we were talking about 120->70. But 120->20? Maybe not.
A dramatic shortening of copyright terms is something of a referendum on whether we want big-budget IP to exist.
In a world of 20 year copyright, we would probably still have the LOTR books, but we probably wouldn’t have the LOTR movies.
> If that asset generates revenue for 120 years, then it’s slightly more valuable than an asset that generates revenue for 119 years, and considerably more valuable than an asset that generates revenue for 20 years.
Not so, because of net present value.
The return from investing in normal stocks is ~10%/year, which is to say ~670% over 20 years, because of compounding interest. Another way of saying this is that $1 in 20 years is worth ~$0.15 today. A dollar in 30 years is worth ~$0.05 today. A dollar in 40 years is worth ~$0.02 today. As a result, if a thing generates the same number of dollars every year, the net present value of the first 20 years is significantly more than the net present value of all the years from 20-120 combined, because money now or soon from now is worth so much more than money a long time from now. And that's assuming the revenue generated would be the same every year forever, when in practice it declines over time.
The reason corporations lobby for copyright term extensions isn't that they care one bit about extended terms for new works. It's because they don't want the works from decades ago to enter the public domain now, and they're lobbying to make the terms longer retroactively. But all of those works were already created and the original terms were sufficient incentive to cause them to be.
Your analysis misses the incredibly important caveat that revenue rises with inflation – or sometimes even faster.
50 years ago, a movie ticket was 0.50 cents in revenue. Today, it’s $25. That’s a 50x increase… a dollar in 50 years might be worth $0.02 today, but a movie ticket in 50 years is worth about a movie ticket today.
> And that's assuming the revenue generated would be the same every year forever, when in practice it declines over time.
For the crown jewel IP that the studios are most interested in protecting, the opposite of this assumption is true. Star Wars, for example, is making more money than ever. Streaming revenues will probably invalidate that assumption for an even wider pool of back catalog properties.
> I’m sure you’re right for individual authors who are driven by a creative spark, but for, say, movies made by large studios, the length of copyright is directly tied to the value of the movie as an asset.
That would be fine, if the studios didn't want to have it both ways. They want to retain full copyright control over their "asset", but they also use Hollywood Accounting [1] to both avoid paying taxes and cheat contributors that have profit-sharing agreements.
If studios declare that they made a loss on producing and releasing something to get a tax break, the copyright term for that work should be reduced to 10 years tops.
Next, they'd switch to from Hollywood Accounting to Oilfield Accounting. Oh that wellhead is actually owned by this other company over there, we just purchased their product at a fair market rate while they were still in business, but now it seems that other company is going bankrupt and cannot do the environmental cleanup to even seal the wellhead, much less remove it.
IIRC, of works that bring in any money to their creators, the vast majority is returned, for almost all works, in the first handful of years after creation. Sure. the big names you know have value longer, but those are a miniscule fraction of works.
Make copyright last for a fixed term of 25 years with optional 10-year renewals up to 95 years on an escalating fee schedule (say, $100k for the first decade and doubling every subsequent decade) and people—and studios—would have essentially the same incentive to create as they do now, and most works would get into the public domain far sooner.
Probably be fewer entirely lost works, as well, if you had firmer deposit requirements for works with extended copyrights (using the revenue from the extensions to fund preservation) with other works entering the public domain soon enough that they were less likely to be lost before that happened.
The Fellowship of the Ring, the first of Peter Jackson's LOTR movies released in 2001, made $887 million in its original theatrical run (on a $93 million budget). It would absolutely still have been made if copyright was only 20 years. And now it would be in the public domain!
The success that we can now measure through hindsight wasn’t assured at the time of greenlighting the film. They took a huge gamble:
https://variety.com/2021/film/news/lord-of-the-rings-peter-j...
It would have been an even bigger gamble if they weren’t able to bank on any long term revenue (I’m certain Netflix continues to pay for the rights to stream the trilogy after 2021).
According to Matt Damon (in one of many interviews) a lot of movies were produced with the second revenue stream (vhs/dvd) being part of the calculations, that is why we now get a lot less variety and alternative movies made, that second revenue stream doesn’t exist any more (I assume streaming pays very little in comparison).
With books it's even worse. Movies might get a trickle of revenue from TV licensing but once a book is out of print (which usually happens very quickly, and most never go into print again), that's it. No more revenue from that book, it continues to circulate in libraries and used bookstores but the author and publisher gets nothing from that.
> If that asset generates revenue for 120 years, then it’s slightly more valuable than an asset that generates revenue for 119 years
Movies for instance make most of their revenue in the 2 week following their release in theater. Beyond, peoole who wanted to see it had already seen it and the others don't care.
I'd argue it's similar for other art form, even for music. The gain at the very end of the copyright lifetime is extremely marginal and doesn't influence spending decision, which is mostly measured on a return basis of at most 10 years.
> If that asset generates revenue for 120 years, then it’s slightly more valuable than an asset that generates revenue for 119 years, and considerably more valuable than an asset that generates revenue for 20 years
Due to the fairy high cost of capital right now, pretty much anything more than 5 years away is irrelevant. 10 years max, even for insanely high returns on investment.
Seems to me most of that inflated budget is needed for the entertainment role of films, not the art in them, which a low budget can often stimulate rather than inhibit. In which case nothing of importance would be lost by a drastic shortening of copyright terms.
OK? So we wouldn't have $100m movies, the vast majority of which are forgotten about in a few months. I don't think a $100m movie is ten times better than a $10m one, so I think I'd be fine with movies with much smaller budgets, if they meant that LotR (the books) are now in the public domain for everyone to enjoy.
Missing is why laws fight so hard too, missing the opposite of what we have (in the west), namely blatant and rampant piracy. The other extreme is really bad, creators of any type pirated by organized crime. There was no video game nor movie market in eastern Europe for example, can't compete against large scale piracy.
Which is to say, preservation without awareness of the threat will look like hoarding. A secondary question is to what extent is that threat real? Without seeing what true rampant piracy looks like, I think it would be easy to be ignorant of the threat.
You're conflating trademark with copyright.
Regardless, it's not just copyright laws that are at issue here. This is reproducing human likenesses - like Harrison Ford's - and integrating them into new works.
So if I want to make an ad for a soap company, and I get an AI to reproduce a likeness of Harrison Ford, does that mean I can use that likeness in my soap commercials without paying him? I can imagine any court asking "how is this not simply laundering someone's likeness through a third party which claims to not have an image / filter / app / artist reproducing my client's likeness?"
All seemingly complicated scams come down to a very basic, obvious, even primitive grift. Someone somewhere in a regulatory capacity is either fooled or paid into accepting that no crime was committed. It's just that simple. This, however, is so glaring that even a child could understand the illegality of it. I'm looking forward to all of Hollywood joining the cause against the rampant abuse of IP by Silicon Valley. I think there are legal grounds here to force all of these models to be taken offline.
Additionally, "guardrails" that prevent 1:1 copies of film stills from being reprinted are clearly not only insufficient, they are evidence that the pirates in this case seek to obscure the nature of their piracy. They are the evidence that generative AI is not much more than a copyright laundering scheme, and the obsession with these guardrails is evidence of conspiracy, not some kind of public good.
> So if I want to make an ad for a soap company, and I get an AI to reproduce a likeness of Harrison Ford, does that mean I can use that likeness in my soap commercials without paying him?
No, you can't! But it shouldn't be the tool that prohibits this. You are not allowed to use existing images of Harrison Ford for your commercial and you also will be sued into oblivion by Disney if you paint a picture of Mickey Mouse advertising your soap, so why should it be any different if an AI painted this for you?
Well, precisely. What then is the AI company's justification for charging money to paint a picture of Harrison Ford to its users?
The justification so far seems to have been loosely based on the idea that derivative artworks are protected as free expression. That argument loses currency if these are not considered derivative but more like highly compressed images in a novel, obfuscated compression format. Layers and layers of neurons holding a copy of Harrison Ford's face is novel, but it's hard to see why it's any different legally than running a JPEG of it through some filters and encoding it in base64. You can't just decode it and use it without attribution.
> you also will be sued into oblivion by Disney if you paint a picture of Mickey Mouse advertising your soap, so why should it be any different if an AI painted this for you?
If the AI prompt was "produce a picture of Micky Mouse", I'd agree with you.
The creators of AI claim their product produces computer-generated images, i.e. generated/created by the computer. Instead it's producing a picture of a real actual person.
If I contract an artist to produce a picture of a person from their imagination, i.e. not a real person, and they produce a picture of Harrison Ford, then yeah I'd say that's on the artist.
> This is reproducing human likenesses - like Harrison Ford's - and integrating them into new works.
The thing is though, there is also a human requesting that. The prompt was chosen specifically to get that result on purpose.
The corporate systems are trying to prevent this, but if you use any of the local models, you don't even have to be coy. Ask it for "photo of Harrison Ford as Indiana Jones" and what do you expect? That's what it's supposed to do. It does what you tell it to do. If you turn your steering wheel to the left, the car goes to the left. It's just a machine. The driver is the one choosing where to go.
No, I think that's unfair. I, as a user, could very reasonably want a parody or knock-off of Indiana Jones. I could want the spelunky protagonist. It's hard to argue that certain prompts the author put into this could be read any other way. But why does Nintendo get a monopoly on plumbers with red hats?
The way AI is coded and trained pushes it constantly towards a bland-predictable mean, but it doesn't HAVE to be that way.
The way it's coded, if you don't specify something more specific, asking for "plumber in a red hat" implies that you're asking for Mario, because that's the most well-known instance of what you requested. If you want something else, you ask for something else. Specify the ways in which you want your plumber in a red hat to differ from the most well-known instance, and it will.
If you want the Spelunky protagonist you can literally type "spelunky protagonist" into the prompt and it will do it, or you can describe how you want your own parody to differ from the original. But if you just type "Indiana Jones" into the prompt and nothing else, you're getting Indiana Jones, because what else is it even supposed to do with that? And likewise if you use a prompt which is designed to conjure India Jones by description rather than by name.
Human appearance does not have enough dimensions to make likeness a viable thing to protect; I don't see how you could do that without say banning Elvis impersonators.
That said:
> I'm looking forward to all of Hollywood joining the cause against the rampant abuse of IP by Silicon Valley.
If you're framing the sides like that, it's pretty clear which I'm on. :)
Interesting you should bring that up:
https://www.calcalistech.com/ctechnews/article/1517ldjmv
Loads of lawsuits have been filed by celebrities and their estates over the unauthorized use of their likeness. And in fact, in 2022, Las Vegas banned Elvis impersonators from performing weddings after a threat from the Presley estate's licensing company:
https://www.dailymail.co.uk/news/article-10872855/Elvis-imag...
But there are also a couple key differences between putting on a costume and acting like Elvis, and using a picture of Elvis to sell soap.
One is that a personal artistic performance could be construed as pastiche or parody. But even more importantly, if there's a financial incentive involved in doing that performance, the financial incentive has to be aligned more with the parody than with drawing an association to the original. In other words, dressing up as Elvis as a joke or an act, or even to sing a song and get paid to perform a wedding is one thing if it's a prank, it's another thing if it's a profession, and yet another thing if it's a mass-marketing advertisement that intends for people to seriously believe that Elvis endorsed this soap.
I can remember two ad campaigns with an Elvis impersonator, and they used multiple people in both of them. I think we can safely assume that if you represent multiple people as a specific public figure, that a reasonable person must assume that none of them are in fact that person.
Now of course that leaves out concerns over how much of advertisement is making money off of unreasonable people, which is a concern Congress occasionally pays attention to.
For what it's worth, this is a uniquely American view of copyright:
> The ONLY reason to have any law prohibiting unlicensed copying of intangible property is to incentivize the creation of intangible property.
In Europe, particularly France, copyright arose for a very different reason: to protect an author's moral rights as the creator of the work. It was seen as immoral to allow someone's work -- their intellectual offspring -- to be meddled with by others without their permission. Your work represents you and your reputation, and for others to redistribute it is an insult to your dignity.
That is why copyrights in Europe started with much longer durations than they did in the United States, and the US has gradually caught up. It is not entirely a Disney effect, but a fundamental difference in the purpose of copyright.
That's an interesting perspective, and yes wholly foreign to my very American economics influenced background.
Are the origins the same when looking at other intellectual property like patents?
How did they deal with quoting and/or critiquing other's ideas? Did they allow limited quotation? What about parody and satire?
I'm not sure about other intellectual property rights. I do know there's a similar dichotomy for privacy: Americans tend to view it in terms of property rights (ownership of your image, data, etc.) while Europeans view it as a matter of protecting personal dignity. That leads to different decisions: for instance, there's a famous French case of an artist making a nude sketch of a portrait subject, and being unable to sell the sketch because it violated the subject's privacy rights, despite being the owner of the intellectual property.
Peter Baldwin's Copyright Wars is a good overview of the European vs American attitudes to copyright in general.
You are missing a bunch of edge cases, and the law is all about edge cases.
An artist who works professionally has family members, family members who are dependent on them.
If they pass young, become popular just before they pass and their extremely popular works are now public domain. Their family sees nothing from their work, that is absolutely being commercialized ( publishing and creation generally spawns two seperate copyrights).
GP's not missing those edge cases; GP recognizes those edge cases are themselves a product of IP laws.
Those laws are effectively attempting to make information behave as physical objects, by giving them simulated "mass" through a rent-seeking structure. The case you describe is where this simulated physical substrate stops behaving like physical substrate, and choice was made to paper over that with extra rules, so that family can inherit and profit from IP of a dead creator, much like they would inherit physical products of a dead craftsman and profit from selling them.
It's a valid question whether or not this is taking things too far, just for the sake of making information conform to rules of markets for physical goods.
There we stand in agreement. Current copyright law and how it is governed is horrible.
What most people propose is equally bad and will never get traction with lawmakers.
Returning to what came before makes a ton of sense. Just make it X number of years and let's debate X for a while to get a decent number.
You seem to talk about fairness. Copyright law isn't supposed to be fair, it's supposed to benefit society. On one side you have the interest of the public to make use of already created work. On the other side is the financial incentive to create such work in the first place.
So the question to ask is whether the artist would have created the work and published it, even knowing that it isn't an insurance to their family in case of their early death.
I don’t know how much time you’ve spent on task scheduling or public strategy, but minmaxing of The Public Good versus the private benefit of art is, in fact, a question of fairness. It’s a compromise to give both parties as much of what they want or need as possible.
Il not sure IP should be used as a life insurance, there’s already many public and private ideas tools for that.
Also it seems you assume inheritance is a good think. Most people do think the same on a personal level, however when we observes the effect on a society the outcome is concentration of wealth on a minority and barriers for wealth hand change === barriers for "American dream".
For writing many people only become popular after they are dead.
I heard this explained once as the art in some writing is explaining how people feel in a situation that is still too new for many to want to pay to have it illustrated to them. But once the newness has passed, and people understand or want to understand, then they enjoy reading about it.
As a personal example, I could enjoy movies about unrequited love before and long after I experienced it firsthand, but not during or for years after. People may not yet have settled feelings about an event until afterward, and not be willing to “pick at the scab”.
The other, more statistical explanation is that it just takes a lot of attempts to capture an idea or feeling and a longer window of time represents more opportunities to hit upon a winning formula. So it’s easier to capture a time and place afterward than during.
>I don't condone or endorse breaking any laws.
Really? Because there are a lot of very stupid laws out there that should absolutely be broken, regularly, flagrantly, by as many people as possible for the sake of making enforcement completely null and pointless. Why write in neutered corporate-speak while commenting on a casual comment thread while also (correctly) pointing out the absurdity of certain laws.
> There are approximately zero people who decide they'll create something if they're protected for 95 years after their death but won't if it's 94 years. I wouldn't be surprised if it was the same for 1 year past death.
I think life of creator + some reasonable approximation of family members life expectancy would make sense. Content creators do create to ensure their family's security in some cases, I would guess.
Trademarks are very different from copyrights. In general, they never expire as long as the fees are paid in the region, and products bearing the mark are still manufactured. Note, legal firms will usually advise people that one can't camp on Trademarks like other areas of IP law.
For example, Aspirin is known as an adult dose of acetylsalicylic acid by almost every consumer, and is Trademarked to prevent some clandestine chemist in their garage making a similarly branded harmful/ineffective substance that damages the Goodwill Bayer earned with customers over decades of business.
Despite popular entertainment folklore, sustainable businesses actually want consumer goodwill associated with their products and services.
While I agree in many WIPO countries copyright law has essentially degenerated into monetized censorship by removing the proof of fiscal damages criteria (like selling works you don't own.) However, Trademarks ensure your corporate mark or product name is not hijacked in local Markets for questionable purposes.
Every castle needs a moat, but I do kind of want to know more about the presidential "Tesler". lol =)
this reminds me of the time I tried to use a prepaid lawyer to research some copyright issues
they went down the rabbit hole on trademark laws, which are not only not copyright related, they are an entirely different federal agency at the Patent Office
gave me a giggle and the last time I used cheapo prepaid lawyers
It is not about trademark laws. It is about where laws apply.
If you torrent a movie right now, you'll be fined in many advanced countries.
But a huge corporation led by a sociopath scrapes the entire internet and builds a product with other people's work ?
Totally fine.
This. It seem the situation is controversial now because of the beloved Studio Ghibli IP, but I want to see the venn diagram of people outrage at this, and the people clamoring for overbearing Disney chatachter protection when the copyright expired and siding with paloworld in the Nintendo lawsuit.
It seem most of the discussion is emotionally loaded, and people lost the plot of both why copyright exists and what copyright protects and are twisting that to protect whatever media they like most.
But one cannot pick and choose where laws apply, and then the better question would be how we plug the gap that let a multibilion Corp get away with distribution at scale and what derivative work and personal use mean in a image generation as a service world, and artist should be very careful in what their wishes are here, because I bet there's a lot of commission work edging on style transfer and personal use.
New rule: you get to keep your belongings for 20 years and that's it. Then they are taken away. Every dollar you made twenty or more years ago, every asset you acquired, gone.
That oughtta be enough to incentivize people to work and build their wealth.
Anything more than that is unnecessary.
I think you mean "20 years after you die". That seems like a perfectly rational way to deal with people who want to be buried in their gold and jewels in a pyramid.
We should never have accepted the term "intellectual property" at all, if this is the mindset it leads to.
Your copyrighted work just has to be an original expression, not derived from someone else's work. It doesn't have to contain original ideas.
Patents are for ideas. Patents do in fact expire far faster than copyrights in the USA. The main problem with patents is patent trolling in the area of software patents.
I disagree that you can own things, but I certainly concede IP as a good starting point.
So you wouldn't mind if I dip my hands into the pockets of the clothing you are wearing and help myself to whatever cash bills I might find?
Also, can you wash this not-my car I'm driving, that happens to be registered in my name?
Let me know if you will be requiring compensation for not-your time and not-your effort.
You will find the needed cleaning materials at the household goods store down the street. Just walk in, grab whatever you need, and walk out.
Not sure if anyone is interested in this story, but I remember at the height of the PokemonGo craze I noticed there were no shirts for the different factions in the game, cant rememebr what they were called but something like Teamread or something. I setup an online shop to just to sell a red shirt with the word on it. The next day my whole shop was taken offline for potential copyright infringement.
What I found surprising is I didnt even have one sale. Somehow someone had notified Nintendo AND my shop had been taken down, to sell merch that didn't even exist for the market and if I remember correctly - also it didnt even have any imagery on it or anything trademarkable - even if it was clearly meant for pokmeonGo fans.
Im not bitter I just found it interesting how quick and ruthless they were. Like bros I didn't even get a chance to make a sale. ( yes and also I dont think I infringed anything).
I asked Sora to turn a random image of my friend and myself into Italian plumbers. Nothing more, just the two words "Italian plumbers". The created picture was not shown to me because it was in violation of OpenAI's content policy. I asked then just to turn the guys on the picture into plumbers, but I asked this in the Italian language. Without me asking for it, Sora put me in an overall and gave me a baseball cap, and my friend another baseball cap. If I asked Sora to put mustache on us, one of us received a red shirt as well, without being asked to. Starting with the same pic, if I asked to put one letter on the baseball caps each - guess, the letters chosen were M and L. These extra guardrails are not really useful with such a strong, built-in bias towards copyright infringement of these image creation tools. Should it mean that with time, Dutch pictures will have to include tulips, Italian plumbers will have to have a uniform with baseball caps with L and M, etc. just not to confuse AI tools?
You (and the article, etc) show what a lot of the "work" in AI is going into at the moment - creating guardrails against creating something that might get them in trouble, and / or customizing weights and prompts under water to generate stuff that isn't the obvious. I'm reminded of when Google's image generator came up and this customization bit them in the ass when they generated a black pope or asian vikings. AI tools don't do what you wish they did, they do what you tell them and what they are taught, and if 99% of their learning set associates Mario with prompts for Italian plumbers, that's what you'll get.
A possible (probably already exists) business is setting up truly balanced learning sets, that is, thousands of unique images that match the idea of an italian plumber, with maybe 1% of Mario. But that won't be nearly as big a learning set as the whole internet is, nor will it be cheap to build it compared to just scraping the internet.
>> they do what you tell them and what they are taught, and if 99% of their learning set associates Mario with prompts for Italian plumbers, that's what you'll get.
I thought that a lot of the issues were the opposite of this, where Google put their thumb on the scale to go against what the prompt asked. Like when someone would ask for a historically accurate picture of a US senator from the 1800s and repeatedly get women and non-white men. The training set for that prompt has to be overwhelmingly white men so I don't think it was just a matter of following the training data.
I would love to know how YouTube does this for music. There's some holes obviously, like some cover artists will play the iconic riffs of a song and then stop somewhere. There's people who do reels or "commentary" of a movie scene and then put some horrible high pitched music to mask it from copyright.
There's probably even some rules around this to only detect just enough to take legal action. Like GP stumbled on a trademark landmine, but obviously just selling red shirts with a bird on it can't be a trademark violation; it needs to be a specific kind of red too.
I remember all the hullaballoo about Asian Vikings and the like. It was so preposterous that Vikings would ever be Asian that it must be ultra-woke DEI mind-worms being forced onto AI! But of course, as far as the AI's concerned, it is even more preposterous that an Italian plumber would not be wearing red or green overalls with a mustache and a lettered baseball cap. I don't see any way you can get the AI to recognize that Vikings "should" be white people and not also think that Italian plumbers "should" look like that. Are they allowed to recombine their training data or must they strictly adhere to only what they've seen?
Of course the irony is that if the people who get offended whenever they see images of non-white people asked for a picture of "Vikings being attacked by Godzilla" , they'd get worked up if any of the Vikings in the picture were Asian (how unrealistic!). It's a made-up universe! The image contains a damn (Asian) Kaiju in it, and everyone is supposed to be pissed because the Vikings are unrealistic!?
I feel the opposite. Like, we just entered the Golden Age.
OpenAI will eventually have competition for GPT 4o image generation.
They'll eventually have open source competition too. And then none of this will matter.
OmniGen is a good start, just woefully undertrained.
The VAR paper is open, from ByteDance, and supposedly the architecture this is based on.
Black Forest Labs isn't going to sit on their laurels. Their entire product offering just became worthless and lost traction. They're going to have to answer this.
I'd put $50 on ByteDance releases an open source version of this in three months.
Many years ago I tried to order a t-shirt with the postscript tiger on the front from Spreadshirt.
It was removed on Copyright claims before I could order one item myself. After some back and forth they restored it for a day and let me buy one item for personal use.
My point is: Doesn't have to be Sony, doesn't have to be a snitch - overzealous anticipatory obedience by the shop might have been enough.
>After some back and forth they restored it for a day and let me buy one item for personal use.
I used Spreadshirt to print a panel from the Tintin comic on a T-shirt, and I had no problem ordering it (it shows Captain Haddock moving through the jungle, swatting away the mosquitoes harassing him, giving himself a big slap on the face, and saying, 'Take that, you filthy beasts!').
Twenty years ago, I worked for Google AdWords as a customer service rep. This was still relatively early days, and all ads still had some level of manual human review.
The big advertisers had all furnished us a list of their trademarks and acceptable domains. Any advertiser trying to use one that wasn’t on the allow-list had their ad removed at review time.
I suspect this could be what happened to you. If the platform you were using has any kind of review process for new shops, you may have run afoul of pre-registered keywords.
Well the teams in Pokemon Go aren't quite as generic as Teamred: they are Team Instinct, Team Mystic, and Team Valor. Presumably Nintendo has trademarks on those phrases, and I’m sure all the big print on demand houses have an API for rights-holders to preemptively submit their trademarks for takedowns.
Nintendo is also famously protective of their IP: to give another anecdote, I just bought one of the emulator handhelds on Aliexpress that are all the rage these days, and while they don't advertise it they usually come preloaded with a buttload or ROMs. Mine did, including a number of Nintendo properties — but nary an Italian plumber to be found. The Nintendo fear runs deep.
It's hard to think of a company more aggressive with their IP than Nintendo
Allen Pan, a youtuber "maker" who runs in the circle of people who run OpenSauce, was a contestant on a Discovery channel show that was trying to force the success of Mythbusters by "finding the next mythbusters!". He lost, but it was formative to him because those people were basically all inspired by the original show.
A couple years ago, he noticed that the merchandise trademark for "Mythbusters" had lapsed, so he bought it. He, now the legal owner of the trademark Mythbusters for apparel, made shirts that used that trademark.
Discovery sent him a cease and desist and threatened to sue. THEY had let the trademark lapse. THEY had lost the right to the trademark, by law. THEY were in the wrong, and a lawyer agreed.
But good fucking luck funding that legal battle. So he relinquished the trademark.
Buy a walrus plushy cause it's funny: https://allen-pan-shop.fourthwall.com/en-usd/
Note the now "Myth Busted" shirts instead.
Hilariously, a friend of Allen Pan's, from the same "Finding the next mythbuster" show; Kyle Hill, is friends enough with Adam Savage to talk to him occasionally, and supposedly the actual Mythbusters themselves were not empathetic to Allen's trademark claim.
> THEY were in the wrong, and a lawyer agreed.
Not sure where you get that from. He doesn't say that in the cease & desist announcement video (though it's worded in a way that lets the viewers speculate that). Also from every time it's brought up on the podcast he's on, it very much seams like he knows that he doesn't have legal ground to stand on.
Just because someone let's a trademark lapse doesn't mean you can rightfully snatch it up with a new registration (as the new registration may be granted in error). It would be a different story if he had bought the trademark rights before them lapsing.
Allen Pan makes entertaining videos, but one shouldn't base ones understanding of how trademarks work based on them.
> my whole shop was taken offline
I think the problem there was being dependent on someone who is a complete pushover, doesn't bother to check for false positives and can kill your business with a single thought.
Yes that was the whole point of my post.
For further info it was Redbubble.
>Redbubble is a significant player in the online print-on-demand marketplace. In fiscal year 2023, it reported having 5 million customers who purchased 4.8 million different designs from 650,000 artists. The platform attracts substantial web traffic, with approximately 30.42 million visits in February 2025.
Idk, the models generating what are basically 1:1 copies of the training data from pretty generic descriptions feels like a severe case of overfitting to me. What use is a generational model that just regurgitates the input?
I feel like the less advanced generations, maybe even because of their limitations in terms of size, were better at coming up with something that at least feels new.
In the end, other than for copyright-washing, why wouldn't I just use the original movie still/photo in the first place?
People like what they already know. When they prompt something and get a realistic looking Indiana Jones, they're probably happy about it.
To me, this article is further proof that LLMs are a form of lossy storage. People attribute special quality to the loss (the image isn't wrong, it's just got different "features" that got inserted) but at this point there's not a lot distinguishing a seed+prompt file+model from a lossy archive of media, be it text or images, and in the future likely video as well.
The craziest thing is that AI seems to have gathered some kind of special status that earlier forms of digital reproduction didn't have (even though those 64kbps MP3s from napster were far from perfect reproductions), probably because now it's done by large corporations rather than individuals.
If we're accepting AI-washing of copyright, we might as well accept pirated movies, as those are re-encoded from original high-resolution originals as well.
The year is 2030.
A new MCU movie is released, its 60 second trailer posted on Youtube, but I don't feel like watching the movie because I got bored after Endgame.
Youtube has very strict anti-scraping techniques now, so I use deep-scrapper to generate the whole trailer from the thumbnail and title.
I use deep-pirate to generate the whole 3 hour movie from the trailer.
I use deep-watcher to summarize the whole movie in a 60 second video.
I watch the video. It doesn't make any sense. I check the Youtube trailer. It's the same video.
Probably the majority of people in the world already "accept pirated movies". It's just that, as ever, nobody asks people what they actually want. Much easier to tell them what to want, anyway.
To a viewer, a human-made work and an AI-generated one both amount to a series of stimuli that someone else made and you have no control over; and when people pay to see a movie, generally they don't do it with the intent to finance the movie company to make more movies -- they do it because they're offered the option to spend a couple hours watching something enjoyable. Who cares where it comes from -- if it reached us, it must be good, right?
The "special status" you speak of is due to AI's constrained ability to recombine familiar elements in novel ways. 64k MP3 artifacts aren't interesting to listen to; while a high-novelty experience such as learning a new culture or a new discipline isn't accessible (and also comes with expectations that passive consumption doesn't have.)
Either way, I wish the world gave people more interesting things to do with their brains than make a money, watch a movies, or some mix of the two with more steps. (But there isn't much of that left -- hence the concept of a "personal life" as reduced to breaking one's own and others' cognitive functioning then spending lifetimes routing around the damage. Positively fascinating /s)
Tried Flux.dev with the same prompts [0] and it seems actually to be a GPT problem. Could be that in GPT the text encoder understands the prompt better and just generates the implied IP, or could be that a diffusion model is just inherently less prone to overfitting than a multimodal transformer model.
[0] https://imgur.com/a/wqrBGRF Image captions are the impled IP, I copied the prompts from the blog post.
If it overfits on the whole internet then it’s like a search engine that returns really relevant results with some lossy side effect.
Recent benchmark on unseen 2025 Math Olympiad shows none of the models can problem solve . They all accidentally or on purpose had prior solutions in the training set.
You probably mean the USAMO 2025 paper. They updated their comparison with Gemini 2.5 Pro, which did get a nontrivial score. That Gemini version was released five days after USAMO, so while it's not entirely impossible for the data to be in its training set, it would seem kind of unlikely.
The claim is that these models are training on data which include the problems and explanations. The fact that the first model trained after the public release of the questions (and crowdsourced answers) performs best is not a counter example, but is expected and supported by the claim.
I was noodling with Gemini 2.5 Pro a couple days ago and it was convinced Donald Trump didn’t win the 2024 election and that he conceded to Kamala Harris so I’m not entirely sure how much weight I’d put behind it.
What if the word "generic" were added to a lot of these image prompts? "generic image of an intergalactic bounty hunter from space" etc.
Certainly there's an aspect of people using the chat interface like they use google: describe xyz to try to surface the name of a movie. Just in this case, we're doing the (less common?) query of: find me the picture I can vaguely describe; but it's a query to a image /generating/ service, not an image search service.
Generic doesn't help. I was using the new image generator to try and make images for my Mutants and Masterminds game (it's basically D&D with superheroes instead of high fantasy), and it refuses to make most things citing that they are too close to existing IP, or that the ideas are dangerous.
So I asked it to make 4 random and generic superheroes. It created Batman, Supergirl, Green Lantern, and Wonder Woman. Then at about 90% finished it deleted the image and said I was violating copyright.
I doubt the model you interact with actually knows why the babysitter model rejects images, but it claims to know why and leads to some funny responses. Here is it's response to me asking for a superhero with a dark bodysuit, a purple cape, a mouse logo on their chest, and a spooky mouse mask on their face.
> I couldn't generate the image you requested because the prompt involved content that may violate policy regarding realistic human-animal hybrid masks in a serious context.
If Indiana Jones (for example) defines a genre, then 'generic' will produce Indiana Jones.
Idk, a couple of the examples might be generic enough that you wouldn't expect a very specific movie character. But most of the prompts make it extremely clear which movie character you would expect to see, and I would argue that the chat bot is working as expected by providing that.
if someone gets Indiana on the suspiciously detailed request the author provided and it appears they wanted something else, they can clarify that to the chat bot, e.g. by copying this your comment.
I have a strong suspicion that many human artists would behave in a way the chat bot did (unless they start asking clarifying questions. Which chatbots should learn to do as well)
To my knowledge this happens when that single frame is overrepresented in its training data. For instance, variations of the same movie poster or screenshot may appear hundreds of times. Then the AI concludes that this is just a unique human cultural artifact, like the Mona Lisa (which I would expect many human artists could also reproduce from memory).
Probably an over-representation in the training data really so it's causing overfitting. Because using training data in amounts right from the Internet it's going to be opinionated on human culture (Bart Simpson is popular so there are lots of images of him, Ori is less well known so there are fewer images). Ideally it should be training 1:1 for everything but that would involve _so_ much work pruning the training data to have a roughly equal effect between categories.
I'm not sure if this is a problem with overfitting. I'm ok with the model knowing what Indiana Jones or the Predator looks like with well remembered details, it just seems that it's generating images from that knowledge in cases where that isn't appropriate.
I wonder if it's a fine tuning issue where people have overly provided archetypes of the thing that they were training towards. That would be the fastest way for the model to learn the idea but it may also mean the model has implicitly learned to provide not just an instance of a thing but a known archetype of a thing. I'm guessing in most RLHF tests archetypes (regardless of IP status) score quite highly.
What I'm kind of concerned about is that these images will persist and will be reinforced by positive feedback. Meaning, an adventurous archeologist will be the same very image, forever. We're entering the epitome of dogmatic ages. (And it will be the same corporate images and narratives, over and over again.)
> I'm ok with the model knowing what Indiana Jones or the Predator looks like with well remembered details,
ClosedAI doesn't seem to be OK with it, because they are explicitly censoring characters of more popular IPs. Presumably as a fig leaf against accusations of theft.
If you define feeding of copyrighted material into a non-human learning machine as theft, then sure. Anything that mitigates legal consequences will be a fig leaf.
The question is "should we define it as such?"
No it doesn't. Commercial intent (usually) makes it illegal, not the fact of copying.
So the criminal party here would be OpenAI, since they are selling access to a service that generates copyright-infringing images.
> I feel like the less advanced generations, maybe even because of their limitations in terms of size, were better at coming up with something that at least feels new.
Ironically that's probably because the errors and flaws in those generations at least made them different from what they were attempting to rip off.
So I train a model to say y=2, and then I ask the model to guess the value of y and it says 2, and you call that overfitting?
Overfitting is if you didn't exactly describe Indiana Jones and then it still gave Indiana Jones.
The prompt didn't exactly describe Indiana Jones though. It left a lot of freedom for the model to make the "archeologist" e.g. female, Asian, put them in a different time period, have them wear a different kind of hat etc.
It didn't though, it just spat out what is basically a 1:1 copy of some Indiana Jones promo shoot. No where did the prompt ask for it to look like Harrison Ford.
But the concentrations of training data because of human culture/popularity of characters/objects means that if I go and give a random person the same description of a character that the AI got and ask "who am I talking about, what do they look like?" there's a very high likelihood that they'll answer "Indiana Jones".
But... the prompt neither forbade Indiana Jones nor did it describe something that excluded Indiana Jones.
If we were playing Charades, just about anyone would have guessed you were describing Indiana Jones.
If you gave a street artist the same prompt, you'd probably get something similar unless you specified something like "... but something different than Indiana Jones".
What would most humans draw when you describe such a well known character by their iconic elements. Think if you deviated and acted a pedant about it people would think you're just trying to prove a point or being obnoxious.
I don't condone or endorse breaking any laws.
That said, trademark laws like life of the author + 95 years are absolutely absurd. The ONLY reason to have any law prohibiting unlicensed copying of intangible property is to incentivize the creation of intangible property. The reasoning being that if you don't allow people to exclude 3rd party copying, then the primary party will assumedly not receive compensation for their creation and they'll never create.
Even in the case where the above is assumed true, the length of time that a protection should be afforded should be no more than the length of time necessary to ensure that creators create.
There are approximately zero people who decide they'll create something if they're protected for 95 years after their death but won't if it's 94 years. I wouldn't be surprised if it was the same for 1 year past death.
For that matter, this argument extends to other criminal penalties, but that's a whole other subject.