Exactly my point that it is not clear, since it’s exactly Carlin’s likeness. A person who tunes in at a random moment has no idea that this is what it it stated in the beginning and could 100% assume it’s Carlin.

It is incredibly clear. The fact that it would take a person to pause the video before the first three seconds, skip to a random point, ignore that the topic of the standup is events that occurred since his death and being an AI, fail to read the written notices on-screen and in the description, etc. is evidence of this.

using their exact likeness as a basis is not transformative work

I think you’re still getting wires crossed between different domains of IP law in a way that makes your objection meaningless. Transformative nature comes in as a part of a fair use defense specifically to copyright infringement - whereas elements of a person’s likeness, like their face or voice, are not protected by copyright.

Do you honestly think that context will matter legally, whether the dead “person” is talking/singing about love or their own death?

Yes, there is legal relevance to whether a reasonable person would interpret the remarks as really being from George Carlin, thus painting him in a false light, and the whole concept of George Carlin riffing on events occurring after his death (plus the disclaimer preceding the video and in the description) is relevant to determining that.

When I say copyright, I mean in a general sence. Infringement of IP might be a better suited phrase, but I assumed the synonymity was implied.

I don’t see how this tracks. Consider your following comment:

You’re either too dumb or stubborn to even google what “transformative work” is. Typical “AI” techbro."

Surely that’s a reference to the character factor of fair use, a defense specifically against copyright infringement? It’s not a term used in trademark law as far as I’m aware for example (and “George Carlin” is not a registered trademark anyway).

Were you just referring to, and telling them to google, the broad layperson definition of “transformative”? In which case I think you’ve misunderstood their comment, because I’m pretty sure at the very least they were referring to the fair use factor.

By selling the bedicked shoe as Nike you are implying that Nike has made this “offensive” shoe and are selling it.

If you do lie to the buyer that it was a brand new Nike shoe, it’d be the concern of the sales contract between you and the buyer, and trademark law.

The Beatles have just officially released a song with their dead singer’s voice.

Lennon’s vocals were recorded before his death, and thus aren’t about his own death and events occurring after it.

No?

To quote the US Copyright office:

Words and short phrases, such as names, titles, and slogans, are uncopyrightable because they contain
an insufficient amount of authorship. The Office will not register individual words or brief combina-
tions of words, even if the word or short phrase is novel, distinctive, or lends itself to a play on words.
Examples of names, titles, or short phrases that do not contain a sufficient amount of creativity
to support a claim in copyright include
The name of an individual (including pseudonyms, pen names, or stage names)
[…]

Go to Spotify and try uploading a track as Michael Jackson, see if copyright “doesn’t protect names or titles.”

I don’t think Spotify allows individuals, as opposed to music distributors, to upload tracks at all - but more importantly their policies on impersonation are not what defines copyright.

Neither example is copyright infringement. The first-sale doctrine allows secondary markets - you are fine by copyright to sell your bedicked shoes to someone.

A sticky note is not a legal disclaimer

Have you watched the video? It’s a thousand times more obvious than any legal disclaimer I’ve ever seen. They are not in any way hiding the fact that it is using AI.

There are many works that talk about death, resurrection, being undead, etc.

Talking about death in the abstract is entirely possible while you’re still alive. Creating material ~two decades after your own death about your death and events that happened since then, less so.

has nothing to do with the title being an obvious infringement.

Copyright doesn’t protect names or titles.

I think it’d be entirely plausible to argue that, while transformative, current generative AI usage often falls short on the other fair use factors.

I don’t really see how it can be argued that the linked example - relatively minor edits to a photograph - are more transformative than generative AI models. What is your criteria here?

The title is “George Carlin: I’m Glad I’m Dead (2024)” and it talks about his own death. Even if someone believes in communication beyond the grave to the extent that they could still mistake it as really being George Carlin, it’s immediately explained as AI in the opening segment of the video.

Whether it’s presented as real seems a reasonable line to me.

Fox News could not use it to mislead people into thinking Biden said something that he did not, but parody like “Sassy Justice” from the South Park creators (using a Trump deepfake) would still be fine.

It’s possible to get away with quite a lot under transformative use even when it’s commercial, consider Cariou v. Prince for example: https://www.artnews.com/art-in-america/features/landmark-copyright-lawsuit-cariou-v-prince-is-settled-59702/

It doesn’t have “memory” of what it has generated previously, other than the current conversation. The answer you get from it won’t be much better than random guessing.

For my understanding artistic works get copyright from the moment of their creation. This would allow one to pick battles based on how lucrative they may potentially be.

In the US, you need your copyright to be registered in order to file an infringement suit or be granted statutory damages. This must be done prior to the infringement, so they wouldn’t be able to pick and choose which to register after the fact. The fact that (unregistered) copyright arises from the moment of creation is true, but not particularly useful here.

You dont really need art museum of babel for this but you just tons of different works that may contain unique characters, structures or objects similar to what someone might be able to imagine or has already imagined.

Copyright is not the same as patents or trademarks; someone coincidentally creating something very similar or even an exact replica of your work is not infringement.

If whether you copied from their work or independently made similar choices is under question - then close similarity of the works could skew the balance of probabilities. However, the courts will be able to see that coincidental similarity is far more likely if a colossal number of images have been registered.

You may draw fan art with disney characters but its actually illegal to sell said art work without Disneys aproval until copyright expires.

It’s still copyright infringement even if you publish it non-commercially, but a Fair Use defense would likely hold up.

To file an infringement suit they’d need to have paid registration for each work which, even for the exorbitantly rich, wouldn’t be remotely feasible for all logical arrangements of words/images. There’s probably not even enough space in the Universe or time until its heat death to generate and store all such images.

Even if they did, copyright doesn’t protect against against independently created works that happen to be similar or even identical - so they wouldn’t be exhausting some limited set of possible works by doing so.

they are looking for specific ones that are in a database

They could be looking for any images without your knowing - there’s no guarantee that those images came from a CSAM database.

Its not like they can create a hash for a guy letting his dog on a horse

They could trivially create a hash for a picture of a guy letting his dog on a horse (which would also include other very similar images).

I didn’t necessarily mean to claim that they can scan for a concept lacking a fixed image, if that’s what you’re saying. That would theoretically be possible with enough hashes, but impractical.

And how was it privacy compromising?

1. Anything could be added to the hashes with the user having no way to know what’s being searched for beyond “trust us”. This could be partially alleviated if, for example, the hash had to be signed by organizations in a combination of states that’d make it difficult to push through hashes for anything other actual CSAM (so not just Five Eyes)

2. Adversarial examples to intentionally set off the filter were demonstrated to be possible. Apple made it clear that there are types of content they’d be legally obligated to report once they became aware of, and it’d be well within a government agency’s capabilities to honeypot, say initially, terrorist recruitment material

3. Coincidental false positives are also entirely possible (ImageNet had some naturally occuring clashes) and can result in their employees seeing your sensitive photographs

4. The user’s device acting against the user cements other user-hostile and privacy-hostile behavior. “People could circumvent the CSAM scan” would be given as another reason against right to repair and ability to see/modify the software your own device is running

5. Tech companies erode privacy by flip-flopping between “sure we’re giving ourselves abusable power, but we’ll stand up to governments pressuring us to expand this” and then “well what were we supposed to do, leave the market?” when they inevitably concede