Tim Boucher

This is a fascinating story that came up as a side-quest in my copyright investigations around 1965, by Zvi Rosen:

The idea of chance in making music is nothing new, the dice game attributed to Mozart is only one of many examples through history. In the early 1950s the composer John Cage pushed this much farther into what would come to be called aleatory music, using randomness system of the I Ching to compose such works dictated by chance as Imaginary Landscape No. 4 for 12 radio receivers, and Music of Changes for piano. In 1952 this led to his best-known and most controversial creation: 4′33″. The “silent piece,” contains no musical notation beyond three movements and an instruction to [be] silent […]

Cage’s publisher Henmar Press applied to register many of them – including 4’33” – with the U.S. Copyright Office, and in looking at how the applications were handled … there’s some valuable lessons to learn about registration of works of indeterminate authorship. Most notably, although a decent number of Cage’s compositions passed muster and were registered as music, others including 4’33” did not, and after years of considering the issue, were registered instead as textual works by the U.S. Copyright Office.

This seems pretty similar to the Sol Lewitt pieces which consist merely of instructions to create a given work, which is then executed by third parties. More to explore down this particular rabbit hole!

I’ve been trying to track down what the heck happened in 1965 in copyright & computers that the Ars Technica piece I referenced earlier places such importance on vis a vis the USCO & computers in art. The Ars piece doesn’t actually say what it was that happened, just this:

But the Copyright Office insisted that the AI copyright debate was settled in 1965 after commercial computer technology started advancing quickly and “difficult questions of authorship” were first raised. That was the first time officials had to ponder how much involvement human creators had in works created using computers.

Skimming through the actual report by the USCO yields not a lot more detail:

Given its role in registering claims to copyright,5 the Copyright Office has considerable experience addressing technological developments related to the creation of works of authorship. As early as 1965, developments in computer technology began to raise “difficult questions of authorship,” including whether material created using technology is “‘written’ by computers” or authored by human creators.6 As then-Register of Copyrights Abraham Kaminstein observed, there is no one-size-fits-all answer:

“The crucial question appears to be whether the “work” is basically one of human authorship, with the computer merely being an assisting instrument, or whether the traditional elements of authorship in the work (literary, artistic, or musical expression or elements of selection, arrangement, etc.) were actually conceived and executed not by man but by a machine.”

This just sounds like more of the “it depends” style of rule-making, which seems still not adequate to me, whatever the year.

Thanks to a ChatGPT conversation though, I did manage to trace down this interesting 2016 paper by Annemarie Birdy, entitled “The Evolution of Authorship: Work Made by Code” (PDF linked at bottom of that page). Some cobbled together quotes from that:

In 1965, Register of Copyrights Abraham Kaminstein identified the question of computer authorship as one of three major problems confronting the Copyright Office. A number of people filed copyright registrations that year for works at least partly authored by computers, and the Office found itself at a loss for how to deal with the situation.

Later:

But does copyright law require human creativity? The Copyright Act doesn’t say anywhere that an author has to human, and there’s really no case law directly on point. Nevertheless, there seems to be an assumption, maybe driven by practical and historical considerations, that authorship means human authorship. The 1965 Register of Copyrights Annual Report [here’s the document] frames the question precisely in terms of the human-computer divide. If a human creates a work, it’s copyrightable. If a machine creates it, then it’s not. The CONTU report does the same thing: unless there’s minimal human creative effort, there’s no protection.

Okay, so this is absolutely mind-blowing that in the actual Copyright Act, there is no requirement that authorship is human-only… The piece talks about a 1956 attempted copyright registration of computer-assisted song-writing:

The only problem that Klein and Bolitho encountered in
their digital composing project was that the Copyright Office refused registration for “Push Button Bertha,” one of Datatron’s many compositions. The reason the Office gave at the time was that no one had ever before tried to register a piece of music written by a machine. The rejection, for which the Office didn’t offer—and couldn’t have offered—any statutory basis, revealed a deep-seated if unspoken
assumption that authors are necessarily human.

Then in a footnote it says this assumption wasn’t codified until 1973:

By 1973, which brought publication of the first Compendium of U.S. Copyright Office Practices, that assumption had become explicit. See U.S. COPYRIGHT OFFICE, COMPENDIUM OF COPYRIGHT OFFICE PRACTICES (FIRST) § 2.8.3 (1st ed. 1973) (stating that works are not copyrightable if they do not “owe their origin to a human agent”).

As a layperson, its difficult to navigate what official standing the COMPENDIUM OF PRACTICES must have relative to the actual law, but it must be… non-zero?

According to ChatGPT – so this must be taken with heaping grains of salt – but it “sounds right” approximately:

The Compendium of U.S. Copyright Office Practices is an administrative manual that outlines the practices and procedures of the U.S. Copyright Office. While it serves as a comprehensive guide for the Office’s staff and the public, it does not have the force of law. However, courts may consider it as persuasive authority due to the Copyright Office’s specialized expertise. 

So…. that’s interesting?

But going back to the Birdy piece, it follows on with an, I think, devastating point (and one which I included in my submission to the USCO originally), that the law already clearly accomodates for “non-human” authorship by way of work-for-hire arrangements, such that a corporation becomes considered legally as the author of the work.

What the anthropocentric view of authorship elides, however, is that copyright law already accommodates a notion of non-human authors; they’re called corporations. Under the work-made-for-hire doctrine, which is a legal fiction, a corporate employer is considered the legal author of a work of which it is not the author-in-fact.38 The statute could have been written to create an assignment by operation of law from an employee-author to her corporate employer, thus maintaining in principle a human monopoly on authorship, but it wasn’t. It was written to allow a corporate employer to be treated ab initio as the author of a work created by its human employee.

Because we already have a copyright doctrine that accommodates non-human authors, maybe that’s a logical place to look for a solution to the problem of computer authors. Maybe we can treat computer-authored works as works made for hire.

So there we are. And that paper is from 2016. So these thoughts have been in the air for quite some time. But still we’re stuck on following a “practices” document, instead of explicitly following the actual law. I’d call that a pretty big discovery for me personally, as these things go…

Continuing to review third party sources talking about the new US Copyright Office report about gen AI + copyrightability, before diving in more deeply into the actual document. This one from Ars Technica has a couple interesting points to bounce off of:

“Very few bright-line rules are possible,” the Copyright Office said, with one obvious exception. Because of “insufficient human control over the expressive elements” of resulting works, “if content is entirely generated by AI, it cannot be protected by copyright.”

Curious what “entirely generated by AI” really means here? I know what I think they probably think it means… something like ‘the AI arranged all the pixels’ (?) maybe? But “entirely” to me speaks more to agency, so to suggest an AI system did something “entirely” on its own would mean something more like on its own agency, without human intervention. Which is not at all the case in task based prompting. Maybe if you had a system that was programmed to randomly put together strings of words and use those as prompts, you could make a bit better case for it to be “entirely” done by the system – rather than more directly by a human prompter in real time. But even then, it’s still initiated, designed, and set in motion as a result of human agency. Well, for now.

But this is why I still think the UK copyright system for computer-generated works is just… better (though I know it’s under review currently… not sure current status). I quoted something here in my statement to the USCO from a UK gov office on their legal situation, which has been on the books since the late 1980s, I believe:

“The “author” of a “computer-generated work” (CGW) is defined as “the person by whom the arrangements necessary for the creation of the work are undertaken”. Protection lasts for 50 years from the date the work is made.”

But anyway, my arguments in that area did not prevail.

Back to the Ars Technica piece:

Following Kaminstein’s advice, officials plan to continue reviewing AI disclosures and weighing, on a case-by-case basis, what parts of each work are AI-authored and which parts are human-authored. Any human-authored expressive element can be copyrighted, the office said, but any aspect of the work deemed to have been generated purely by AI cannot.

“Deemed” how, exactly? This will be one thing I am on the lookout for as I begin my deeper dive into the USCO report. Because it may not be altogether very obvious at times – in my books, for example (even I don’t know a lot of the time looking back on some) – where the human ends and the AI begins in creative collaborations with these tools. And unless there is some foolproof method that’s being proposed here… well, we might be back where we started.

“Where a human inputs their own copyrightable work and that work is perceptible in the output, they will be the author of at least that portion of the output,” the guidelines said.

Again “perceptible” sounds a bit wiggly in terms of interpretation. I’m feeling like a lot of these things are not enforceable unless we take works in question into Photoshop and zoom way in at a pixel-by-pixel level to decide which pixels are attributable to a human and which to a machine. That might not be the intent, but it sounds tough to scale for copyright application assessors. For artists, I guess they could start separating out layers of AI versus human actions? Does C2PA or Content Credentials work at this micro-level? Do we have to send our full file history with copyright applications now?

Excellent video from a year ago:

It’s a “high level” art think piece, not just a bunch of anime gen AI images or whatever. Capital A Art.

I’m still sifting through the new AI + copyrightability report put out by the USCO, the one my submission was cited in regarding Jackson Pollack. Have been combing like articles, like the one from Artnews cited below, to try to get a foothold into it before continuing to comb through the document line by line.

The report illustrated how easily matters of authorship can be muddled with AI using the example of a Gemini-generated cat smoking a pipe and reading a newspaper. Gemini, the reported noted, ignored select prompt instructions and added elements of its own, including the cat’s “incongruous human hand.”

The unpredictability of Gemini was then contrasted with examples of human spontaneity, like the splatter technique of Jackson Pollock. The artist did not manage where or how the paint landed, but “controlled the choice of colors, number of layers, depth of texture, placement of each addition to the overall composition — and used his own body movements to execute each of these choices.”

“The issue is the degree of human control, rather than the predictability of the outcome,” the office concluded.

However, the department said that using such technology to assist in “human” creative expression does not necessarily preclude a work’s eligibility for copyright protection. Like a writer asking AI to create an outline for a book, the user is “referencing, but not incorporating, the output,” the report explained.

To me, when I read that as an artist, I almost take it as a challenge. My brain starts formulating “Oh yeah? Well, what if I ______ and used ____ to ____?” Like literally every criteria they mention above, I can think of half a dozen interesting ways to tear it all down and remake it into something new… And speedrunning through videos on historical art movements has given me tons of new (well, old, since most of its historical – but new to me) references and art historical contexts to flesh out my understanding of how we arrived at this moment, and what relevant things artists have already said about these issues many times over during at least the last 125 years, perhaps longer. More on that another time though, as those project ideas simmer on the back of the stove.

One thing I’ve spotted in media coverage around that report has been that in terms of purely reading headlines about it in Google News, it’s about a 50/50 split, it seems, as to whether or not the position championed by the Copyright Office enables something (like copyrighting art that is assisted by but not purely generated by AI), or restricts something (like registering copyright for, I guess, purely AI generated works). So I wonder if that is a success or failure in terms of public messaging? I would guess maybe it shows success in finding some middle position that much of the public is likely to be able to live with… we shall see.

Decent short video about the process of moving a Sol Lewitt piece in a museum:

Some context from his wiki page:

According to the principle of his work, LeWitt’s wall drawings are usually executed by people other than the artist himself. Even after his death, people are still making these drawings. He would therefore eventually use teams of assistants to create such works. Writing about making wall drawings, LeWitt himself observed in 1971 that “each person draws a line differently and each person understands words differently”. Between 1968 and his death in 2007, LeWitt created more than 1,270 wall drawings. The wall drawings, executed on-site, generally exist for the duration of an exhibition; they are then destroyed, giving the work in its physical form an ephemeral quality. They can be installed, removed, and then reinstalled in another location, as many times as required for exhibition purposes. When transferred to another location, the number of walls can change only by ensuring that the proportions of the original diagram are retained.

There’s a famous quote that is attributed to Lewitt, and which appears to originate in this 1967 Artforum piece, Paragraphs on Conceptual Art, which I’ll quote just a short version of here:

When an artist uses a conceptual form of art, it means that all of the planning and decisions are made beforehand and the execution is a perfunctory affair. The idea becomes a machine that makes the art.

Presumably, that article is an extension of his original piece ‘Sentences on Conceptual Art,’ the text from which is visible here. I’m trying to figure out how Sol Lewitt’s artwork presages some of what we see in generative & algorithmic art today… Are Lewitt’s full-scale wall drawings that are executed by crews of artists & craftspeople considered copyrighted to him?

Like any good conceptual artist, I asked ChatGPT for its opinion, and it gave me this:

Sol LeWitt’s conceptual art practice directly parallels the AI copyright debate, particularly regarding authorship, execution, and human intervention. LeWitt created instruction-based artworks, where he provided precise, often algorithmic guidelines for others to execute his pieces—sometimes without his direct involvement. Despite this, LeWitt was still granted copyright over these works because the creative idea and system of execution were his intellectual property, even if the physical realization was carried out by others.

This raises key contradictions in AI-generated art:

• LeWitt’s logic applies to AI prompting: If a human provides structured prompts or generative rules, why is the resulting image considered AI-authored rather than human-authored?

• Does execution matter? If human assistants executing LeWitt’s instructions didn’t negate his copyright, why does AI “executing” a prompt remove authorship?

• Minimal human intervention: LeWitt’s wall drawings are often executed by fabricators with no creative input of their own—similar to an AI merely realizing pre-set human parameters.

If LeWitt’s purely conceptual authorship was enough for copyright, then AI-assisted art following structured human direction should logically qualify as well. The US Copyright Office’s position contradicts the art historical precedent set by LeWitt’s work.

Like many things ChatGPT, I’m not sure how accurate that all is in the specifics. It depends heavily on the assumption that Lewitt’s third-party produced drawings were/are copyrightable to/by/for him, something I have not yet been able to independently establish proof of.

I did find someone asking some related questions here on Reddit around Lewitt’s work and copyright. They point to the practice of Lewitt’s of issuing certificates of authenticity for a given work, a copy of one of which is presumably viewable here:

I haven’t dug too deep or verified much of this outside of the above fragments, as I’m during a speedrun through art history, but this is a good follow-on:

Free Sol LeWitt is a workshop producing copies of the work by Sol LeWitt, Untitled (Wall Structure), 1972. The workshop produces exact replicas of the artwork in metal and distributes them to the public, free of charge. 

LeWitt famously wrote that “I believe that ideas, once expressed, become the common property of all. They are invalid if not used, they can only be given away and not stolen.” Taking the artist’s word seriously, Free Sol LeWitt questions the restrictions of copyright, the relevance of originality, and the tendency to safeguard art in museums, distant from people’s daily lives. Through a lottery system, the works are “set free” from the museum or gallery setting, to take up residence in people’s homes.

Separately, this blog post seems to contradict ChatGPT’s ‘confidently wrong’ assertions above:

The international TRIPS (Trade Related Aspects of Intellectual Property Rights) Agreement—the treaty with the most nation signatories—puts the matter perhaps most succinctly: “Copyright protection shall extend to expressions and not to ideas, procedures, methods of operation or mathematical concepts as such.” The expression is the formal thing—the words, the shapes, the sounds. This is the thing(to use the language of the U.S. Copyright Act) that is “fixed in a tangible medium of expression.” This, so far as the law is concerned, is the thing that is protected: those shapes and those colors in that arrangement. Insofar as Erased de Kooning Drawing is a work of conceptual art—and insofar as a work of conceptual art is not a final product but a process, a project, an idea—it isn’t protected by copyright. What is protected is what I like to call the authored work—the more impoverished thing with which copyright concerns itself: the merely formal thing—the visual composition, the arrangement of shapes and colors, the end result.

Copyright might also protect documentation of the work’s creation or instantiation, or the instructions for making that work. Where the construction of such a work is photographed or filmed, the photograph or film would be protected by copyright in the same way that, say, the Zapruder film of JFK’s assassination is protected: as the capturing of some moment in history, some fact from some perspective. The fact is not protected, only the document. The instructions for instantiating such a work, meanwhile—like Sol LeWitt’s wall-drawing instructions—would be protected in the same way that a recipe or the instructions for a game are protected, which is to say “thinly”. Those precise words in that precise order would be protected against illicit copying, but the method, system, or process they describe would not.

Anyway, a good enough place to end this ramble for tonight…