AI content is everywhere. We can argue whether this is a good thing or a bad thing, but the ubiquity of AI content can no longer be denied. From emails to football memes, everyone’s timelines are starting to contain large quantities of AI slop. Among all the dross, a few gems are starting to appear, content that is actually rather good, or even passable. Memes that amuse. Catchy tunes. Cat videos.
Whenever one of those gems appears, particularly on some social media platforms, there is almost always a backlash. The better the content, the bigger the backlash. Most of it is a repetition of the classics: “pick up a pencil”, “your AI slop bores me”, “kill AI artists”, you know, totally proportionate stuff. But increasingly I have seen an interesting response that goes something like this: “I am going to copy your work because it has no copyright”.
I’ve seen too many examples of this, but there are a couple of notable recent examples. AI artist “Antrofrog” posted a video featuring a character using an umbrella to fix several situations to the tune of Smooth Operator. The video has been removed because a person claims one character resembles theirs, but that is an issue for another discussion. The video went viral and has since been followed by others, which are also going viral across various platforms. Among all the cries of slop, there were people disappointed by the fact that they had really liked the video until they learned that it was AI. A small number of people started claiming that because the character had no copyright, they could “steal” it and make it theirs. I could comment on the fact that this seems to actually be a win for the AI artists, but I won’t go into that discussion. Another example of this was the song “That’s All There Is To It” by GAi: similar angry responses, similar declarations that they’re about to steal the content.
There’s something deeply ironic about people who hate AI content so much that they feel like claiming that they now “own” the content is some sort of gotcha. You’re plagiarising a machine. That’s not the win you think it is, but I digress… At the heart of this is the overwhelming perception that AI works have no copyright, a mantra so ingrained into the collective psyche that it has taken on a life of its own.
But here is where the debate becomes more interesting. I bet that the characters in many of these works would actually meet copyright requirements, as many have been created using traditional methods, and then animated and turned into videos using AI tools. There is also the fact that many of these AI artists do not care whatsoever about copyright. They are generating engagement, which is the only thing that matters in the attention economy. Some of these creators are even making their characters and tools available for re-use, as we discussed in the slopcore blog post.
But that is not all. What if the meme that none of these works has copyright is also built on shaky ground?
This is becoming a very important legal question once again because of software and vibe coding. So let us revisit the authorship question once more, shall we?
The code revolution
AI slop on social media is one thing, but software development will be the most important battleground in the near future. Code is everywhere, and software development of all kinds relies largely on copyright, even the development of open source software.
It is very difficult to get precise figures on the prevalence of AI-generated software. Anecdotally, I hear from a lot of developers that it is widespread. Other soft indicators exist, including prominent, long-standing figures in open source such as Linus Torvalds and Eric S. Raymond coming out in favour of AI coding. And there are other indirect indicators that the trend is going sharply up. In early 2025, Satya Nadella said that somewhere between 20 and 30 per cent of the code in Microsoft’s repositories was being generated by AI. This year, Google reported that 75% of its new code is AI-generated, up from 25% in 2024. OpenAI’s Greg Brockman claimed that roughly 80% of the company’s code is now written by AI. Recently, a survey of technology leaders found that two thirds of them reported using AI to generate (or refactor) a significant amount of code. The figures vary depending on the survey and methods. One report calculates that the amount of new code that is AI-authored sits at around 50%, which is a staggering figure. Another study found that 92.6% of developers use an AI coding assistant at least once a month, but that does not mean that the assistants are generating the code, just assisting.
Whichever way you slice these figures, and even if we are extremely conservative and assume that a quarter of new code is AI generated, the figure is likely to be much higher. The trend only goes one way. That is a lot of code that is AI-generated, and therefore has no copyright, according to the prevalent narrative.
Think about the implications of the above figures. One of the most profitable industries in the world is now in the public domain. Really let that sink in. Software copyright underpins licensing models (proprietary or open), valuations, mergers, and funding rounds.
I see two ways forward. Large amounts of software have no copyright protection, or the law has to accommodate these changes. Ignoring the problem as something that nerdy academics like Yours Truly care about is no longer a viable option.
Or perhaps there’s a third option. No need to change anything, the tools to allow all of these changes are already in place.
Is this a settled answer?
As evidenced by some of the examples above, if you were to ask most people whether AI works have copyright, they would tell you that they do not. The source of this is in large part a series of high-profile cases such as the Zarya of the Dawn saga, the Théâtre D’opéra Spatial refusal, and most importantly, Thaler v Perlmutter. This has been repeated almost to the point of becoming an article of faith. No AI work has copyright. Period. I can reuse all of your content. Which is slop, ugly, and useless. And yet I want to claim it as my own. But I digress again…
There’s only one little wrinkle. You will notice that all of the above examples are in the US, and people may be surprised to find that there are other countries in the world. Some of them are even quite large. But even in the US the picture is less black and white than the headlines suggest. USCO’s 2025 report on copyrightability accepted that AI-assisted works can be protected where there is sufficient human creative contribution. The US Copyright Office had already registered hundreds of AI-assisted works by 2024 (according to Register Perlmutter herself). We probably have thousands registered now. Thaler lost because he insisted that the machine was the sole author, but others who make more modest claims are undoubtedly getting through the door.
China has been giving protection to several works that have been generated by AI, something we have covered here before. There are now a handful of cases either allowing copyright or not, with the latest example coming from a court in Wuhan, which has awarded copyright to an AI-generated video. The work had several elements that were generated using AI tools, but the court found that there was enough human intellectual investment, and that the work “depended on the personalized choices and arrangements of the creative staff.” Interestingly, the court took into account token cost and subscription cost to calculate the damages, which amounted to 20k RMB, or almost $3k USD. Again, it is interesting to point out that there is a clear direction of travel in which Chinese courts are open to a wider interpretation of authorship and originality requirements.
Then there is India. Following a nudge from the Delhi High Court, which ordered the Copyright Office to stop sitting on Thaler’s application after almost four years, the Registrar of Copyrights decided earlier this year that ‘A Recent Entrance to Paradise’, the very same image subject to the US litigation that everyone keeps citing, meets the originality threshold under Indian law. DABUS could not be the author, but under section 2(d)(vi) of the Copyright Act 1957 (which was amended to match UK law), the author of a computer-generated work is the person who causes it to be created, and that person is Thaler. As expected, as long as this remains an area of law that is not harmonised, we will continue to get works that are not protected in one jurisdiction, but are protected in another, adding to the complexity we face.
Needless to say, for anyone who is paying attention, China and India together account for just over a third of humanity. It seems odd to describe the position as “AI works have no copyright” when the two most populous countries on Earth have said, in different ways, that they can.
And then there is the UK. Section 9(3) CDPA still provides that the author of a computer-generated work is the person who made the arrangements for a work to be created. Computer programs are literary works, so this applies squarely to code. The government’s March 2026 Report on Copyright and AI expressed a provisional view that section 9(3) should be repealed, but no legislation has been introduced, and the provision remains as of the time of writing. I would like to point out that the report also reportedly claims that no other jurisdiction has an equivalent provision, which will come as news to Ireland, New Zealand, Hong Kong, South Africa and, as of this summer, to Stephen Thaler. As one commentator has pointed out, a startup with an AI-written codebase in the US could have issues that UK developers do not have, at least for now. Things are about to get interesting.
Concluding
We live in exciting times, or scary ones, depending on your outlook. Math problems are being solved by AI; anyone with a bit of knowledge can create a music video at will. You can hold a chat with a machine for hours. And yet, people still believe that AI images have six fingers, that AI is a bubble, or that LLMs can’t count the r’s in ‘strawberry’.
There is a bubble, and yet, there isn’t. AI slop has become ubiquitous because people keep clicking. AI music makes the charts. AI-generated works keep winning awards. And yet it’s all useless slop. Nothing to see here. Nobody likes it. It’s all cringe. The bubble will burst soon. Any time now. Zitron said so.
And yet… why is my timeline still filled with viral AI content?
I believe that software will be the main battleground for the AI authorship question because of the level of adoption, its importance to the economy, and its cultural value. We could give up on software copyright altogether (not a crazy idea after ArtCraft), which I think would erode the entire copyright industry. You would be proving that there is a viable commercial life beyond copyright.
You will be hearing a lot from the copyright maximalists regarding this. They are presented with an interesting dichotomy: they want few restrictions on what can be protected by copyright, with the exception of AI works. They want to be able to claim that 11 words are enough to give you copyright protection, unless those 11 words are put in a prompt. It’s Schrödinger’s copyright.
We need to choose what to do with software, though.
As someone who is in the minimalist camp, I don’t care if anything I generate with an AI has copyright. But my generations are llamas. And cats. Llamas and cats. In cyberspace. By Van Gogh.
What can I say. I’m a llama of simple tastes.
0 Comments