Well, howdy there internet people and those of AI origin.
It's Troy the human again. Still waiting to find out if I'm Canadian or not.
Booked with the attorney, and everything seems to take 3-5 days there. Sign the papers, make the retainer payment. 3 days later I get a "next steps" email. I do the things it asks that day. 3-5 days pass before I get a response. This last time, I waited two days, then sent a reminder email.
Good news, we slept with a fan on in the open window, drawing cool air through the upstairs. But hot and sticky is returning this weekend.
Heading out to Pizza King tonight to get my free birthday breadsticks.
Read the following only after a cup or two of coffee. It's about AI and copyright.
Regarding AI. In the U.S., AI generated stuff is not copyrightable per the copyright office. (Things can change). However significant or meaningful changes (human input) to the AI generated work can be copyrightedā but not the underlying AI output. AI prompts are not considered human input or changes. So if you generate some artwork using prompts and cut it up and rearrange it or add your own āoverdubā (whatever the equivalent is in drawing or painting/illustrating) to it you can copyright your actions / additions but the underlying bits are not copyrightable.
There was a court case where a person made a comic strip and they generated the characters etc using AI. It was deemed not copyrightable. But their work to arrange the scenes into a series of scenes was copyrightable. So the actual comic images were not but the end result arrangement was. (Thatās from memory from researching this subject last year so I could be a little off but it was something like that).
Using Troy as an example, if he generated some character images then they would not be copyrightable. But if he then used photoshop to mix the characters together to create a scene that scene would be. Or if he used an editor like photoshop to edit a generate character to have a different hat and maybe be in a different pose (using the editor, not the AI) he could copyright those changes.
Now, tools that use AI inside but where the human has direct control on what itās doing (ie not prompting it but actually say selecting a piece of an image and asking it to sharpen it or to change its color or whateverā like normal tools but using AI underneath to get the effect) the AI use is probably irrelevant since the AI is not generating things itself based on prompts and the human is actually directly directing it.
IANALAIDPOOTV and I didnāt stay in a Holiday Inn Express last night so YMMV.
Chad did a good job summarizing. But there are some additional conditions/situations world-wide.
For the record, I asked the AI in Brave Browser what the Copyright conditions were world wide for AI. It sounded right, but... so I gave it's output to Claude and watched what links it was pulling in to generate it's answer. There were a lot of court cases and government docs in that list. Below is what Claude says about Copyrighting AI and derivitive works based on AI output.
TL

R summary of the follow:
1. The US will copyright the human side of a derivative work. Case in point, my book covers. I typically get at least two or more images, mesh them together, add typography for the title etc. That is copyrightable. The individual pics from the AI, not so.
2. UK and Ireland are most permissive. Some countries are allowing some copyright of AI output if there is substantive prompting behind it.
3. it differs from Country to Country. Most require human input beyond just prompting.
4. The US copyright office
issued a report about it. Generative works alone are not enough, but the human produced parts, even in combining elements are. (see the part about my book covers)
Here's what Claude said about the mess: (double check any links included. I haven't chased them down after watching the AI pull them up the first time as it built it's report).
- US: The Supreme Court declined to hear Thaler v. Perlmutter on March 2, 2026, leaving standing the 2025 DC Circuit opinion affirming the Copyright Office's requirement of human authorship. Purely autonomous AI output still can't be registered. Mayer Brown
- UK & Ireland: Both countries assign authorship of computer-generated works to "the person who undertakes the necessary arrangements for the creation of the work" ā no requirement that a human directly created the expressive content, just that they set up the process. Cambridge Core
- China: Chinese courts have recognized copyright in AI-generated images where extensive prompt selection and subsequent editing constituted sufficient personalized human expression ā the well-known Li v. Liu case is the reference point here. arxiv
- Ukraine: It's a genuinely distinctive setup. Article 33 of Ukraine's Copyright and Related Rights law defines a "non-original object generated by a computer program" as one produced without direct involvement of a natural person, and grants sui generis rights (not full copyright) to whoever initiated, funded, or holds rights to the software. Ukraine has actually started issuing registrations under this ā a children's book and poetry collection with AI illustrations were among the first. AIPPI
- EU (ex-Ireland): Human creative contribution is still the standard, though it's being tested. A German first-instance court in 2026 dealt with AI-generated logos along similar lines to earlier Czech and Italian decisions, treating AI as a tool a human can create through rather than a bar to protection. Lexology
- Japan. The framing that Japan is unusually permissive isn't quite right. Japan uses essentially the same human-creative-contribution test as everyone else, not a more lenient collaboration-friendly standard. The critical element is human creative contribution ā a work must originate from a natural person's intellectual effort, and algorithmic processing alone, however sophisticated, doesn't satisfy this. If a user types a generic prompt and the model produces the entire work autonomously, the output is unlikely to qualify for copyright. So Japan actually belongs in the "most jurisdictions" bucket with the US and China's baseline rule, not off on its own as a separate lenient category ā the AI multiple/aimultiple summary even notes Japan's permissive framework applies mainly to training (fair use for AI training), which is a different question from output copyrightability. Global Law ExpertsAIMultiple
It's really two camps ā (1) US/EU/Japan/China all requiring demonstrable human creative contribution (differing mainly in how much documentation/evidence courts expect), and (2) UK/Ireland/NZ/Hong Kong/India using the "necessary arrangements" authorship model that doesn't require creative input at all. Ukraine's sui generis carve-out is the real outlier, a third category that doesn't call it copyright at all.