Owning an original piece of art doesn't mean you have the copyright. The most the guy with the cels could do would be to publicly display them. This is the same sort of foolish behavior that the dopes that bought that Dune book the other year engaged in.

Two separate things, really.

From a copyright perspective, all that is in the public domain are the first three Mickey Mouse films. Attributes of the character in those films -- appearance, behavior, etc. -- may be used. Copying other attributes from still-copyrighted sources is not permitted, such as the particular voice, the color appearance of the character, etc. As those sources drop into the public domain too, over the years, they'll also become available.

But from a trademark perspective, the issue is whether customers may be confused as to the origin of goods bearing a Mickey Mouse logo (either as a character or even the word mark) -- are they from Disney or somewhere else? Also, whether junior users of the mark are diluting the uniqueness of the unquestionably famous Mickey Mouse mark even where there is no confusion by customers. This is typically framed as either blurring -- associating the mark with goods it's not normally applied to -- and tarnishment -- weakening the mark by associating it with unsavory, unflattering, or inferior goods.

So when you say:

They've already done this. (Also, even if they hadn't, Disney isn't losing rights to use Mickey Mouse themselves -- they're just losing rights to control others' use of it)

This brings us to the last couple important concepts: 1) Trademark is not a substitute for copyright; 2) When a copyright or patent enters the public domain, it can have the effect of causing a trademark on the thing that is now in the public domain to be lost through becoming generic.

Let's tackle the second one first: The point of a trademark is to give consumers assurances that goods marked with the mark share a common origin and are of the same level of quality (whether good or bad is irrelevant; the point is consistency). So if you see a bottle marked COCA-COLA, you know it must originate from the Coca-Cola Company and that it's going to taste like any other Coke. If Pepsi starts to sell soft drinks labeled as COCA-COLA but containing Pepsi, this is infringing because it confuses customers: they don't know where a given marked bottle comes from, or what's in it. The Coca-Cola Company can sue to stop this. But if they failed to protect their rights, they'd lose them because they would be allowing confusion to run rampant, and confusion is incompatible with a viable trademark.

In the late 19th century, Henry Perky invented shredded wheat and got a patent on the food itself, and on the method of making it. Since no one else could make it or sell it due to the patent, SHREDDED WHEAT functioned as a trademark; anything marked that way must come from Perky (or Nabisco, who ultimately acquired the rights), and would be consistent in quality. When the patent expired, rival company Kellog's started making it too, and also used the SHREDDED WHEAT mark. The US Supreme Court held that this was not infringing because with the expiration of the patent, everyone was free to make it, and call it by its name. Thus, the loss of the patent killed the trademark.

In a much later trademark case, the Court also held that trademark could not substitute for copyright; once a work is in the public domain, trademark cannot be used to prevent the public from using it.

Together this means that anyone can start making new Mickey Mouse films derivative of the public domain ones, starting next year, and using the Mickey Mouse character by name. This will kill the trademark, at least in the realm of creative works -- books, movies, comics, etc. Maybe sculptural works too, like plush toys (since the Mickey Mouse shape is the point, rather than a source identifier -- think of the difference between a shirt with a picture of a polo player and a shirt with the POLO mark on it). But not non-creative works, like Mickey Mouse-branded ice cream bars or wrist watches.

Certainly there's nothing that prevents the use of public domain characters as trademarks, like PETER PAN branded peanut butter, or KING ARTHUR branded flour. But equally, those trademarks don't prevent anyone from making new books or movies or what have you about the characters of Peter Pan or King Arthur.

All that being said, Disney has loads of money and is famously litigious, so it'll be interesting to see who goes first to test the waters.

Well, you can create derivatives based on the original version. Want him to have lines of dialog? That's fine, just don't copy the high pitched voice that was introduced in later, still protected cartoons. Instead have him sound like Barry White. Want him to have a different look? That's fine too, just don't copy from changes to the look introduced in later cartoons -- blue shorts instead of red (or the rarely seen green), for example.

It's been done before in the US.

Not only no, but hell no. It should be massively shortened.

I agree generally, except as to the last point: the character is Mickey Mouse. Nominative use permits the use of his name. In fact, given the almost total lack of dialogue or title cards for Steamboat Willie, and given the marketing of the thing ('Mickey Mouse in Steamboat Willie'), there's no reason to think that that is even the name of the character the Mickey Mouse character is portraying. Steamboat Willie is merely the name of the cartoon, and is evocative of Steamboat Bill, which is the pre-existing song that Mickey is whistling at the beginning.

In addition, remember that there are actually three Mickey Mouse short films hitting the public domain in 2024: Steamboat Willie, Plane Crazy, and The Galloping Gaucho. He can't be all of them, and Plane Crazy doesn't even make sense as a character name.

It depends on how you use it. If you use the Mickey Mouse character in a new creative work, that should be fine. If you use the Mickey Mouse character in connection with goods or services otherwise (e.g. on the wrapper of Mickey Mouse-branded chewing gum, for example), then you'll probably have legal trouble.

No it didn't, that's just stupid.

There are simply several different terms for different circumstances. You can find them easily at 17 USC 302-305.

For works created on or after January 1, 1978, the term is the life of the author plus 70 years. (Originally it was life + 50, and the extension simply added 20 years)

However, if the work is a joint work by multiple authors, the life in question is the life of the last surviving author.

And if the work is a work made for hire, or psuedonymous, or anonymous, then the term is a straightforward 95 years from the date of its first publication, or 120 years from the date of its creation, whichever is shorter. Life is not a factor whatsoever, unless the author reveals their identity by certain means before the time period runs out, in which case it switches over to the usual life + 70 rule as given above because now it's practical to determine that. Again, the terms had been 75 or 100 years, and the extension simply added 20 years.

There are additional rules for works created before but not published by January 1, 1978, and for works which were already copyrighted prior to January 1, 1978.

There's several, but the one relevant to this discussion is fair use, at 17 USC 107. While the for-profit nature of the use is a factor to be considered in the fair use analysis, that does not, by itself, indicate that the use is unfair.

I would suggest taking a look at Author's Guild, Inc. v. Google, Inc., 804 F.3d 202 (2nd Cir. 2013) for a similar project in which a for-profit corporation copied and used copyrighted materials in their own project, in bulk, for profit, and was held not to have infringed on copyright in the process.

Given that AI training is not some sort of magic compression that breaks Shannon's Law, the use of works to train software is, if anything, even more fair than Google Book Search, which does copy entire works and store them permaently, even if they are not fully made available to the user.

These things are always slower and harder to support than just a standard workstation, support and ongoing costs are higher. 2GB memory was ok for tiny low cost devices 10 years ago, now it's just a joke.

It's not impossible to configure a set of applications that will work in such a poor spec, but most teams can't or won't do it. I think a stop-gap solution might be in order, want to run a software stack with little CPU, a tiny allocation of ram and no local storage? Try it on a physical machine that can be scaled back up.

It probably doesn't even need to be tried at all though, just put an actual user on one of your test environment machines and see how many minutes it takes before they run into a work stoppage.

Commonly tethering is detected by a different ttl, as the connected device will have an extra hop on the way to the carrier router. Some software can be had to alter this count which may evade detection.

This is pretty much what everyone expected. I was always surprised that Apple didn't give themselves a workaround by offering -- at least for the very high end -- the option to put new guts into the old case. Not really their usual MO, but it might have helped sell some more of the super expensive models, if that was important to them.

Copyrights in a character is really just a shorthand way of referring to the portions of one or more copyrighted works that describe the character, often as a gestalt.

But there is not really such a thing as a character copyright; only copyrights in works.

If you make your own Mickey cartoon you are at least infringing on the Steamboat Willie copyright from 1928 and maybe more depending on what version of Mickey you use.

Good news though: The 1928 version hits the public domain in three months. So long as you don't copy from later, still copyrighted works, you will be able to go nuts.

(For example, his voice came out in 1930, as I recall, so until the work that introduced the distinctive Mickey voice is also public domain, better err on the side of a Mickey that sounds like Barry White)

You said two ways and first -- what's the second?

Also, the person who holds the copyright of a preexisting work used as the basis for a derivative work does not get the copyright of an infringing derived work. That would be unconstitutional, and it's not in the law. Rather the copyright of a derivative work is only on the original portions, and to the extent that it is derivative an unauthorized work may simply not be copyrightable. This is all under 17 USC 103.

We're basically on our way to making something like the Star Trek holodeck -- you give the computer some instructions as to what you want, as basic or detailed as you like, and then you have fun with it.

Frankly, I'm looking forward to it.

Ugh, no.

The "legal term for a copyright violation is literally" copyright infringement. You can see this at 17 USC 501.

Here is the definition in the law of 'derivative work':

While it can be an infringement to prepare a derivative work under 17 USC 106, it can also be perfectly legal -- such as if the preexisting work is in the public domain, is used by that work's copyright holder or with their permission, or if it falls under some other exception to copyright, such as fair use under 17 USC 107.

For example, Detective Comics 27 introduced Batman. Pretty certainly every subsequent Batman comic is ultimately derivative of that. The Batman films and tv shows are derivative works based on the comics, etc.

Nothing wrong with a work being a derivative work. They can even be quite good, like how the Aeneid is fanfic of the Iliad.