If we had a policy of nationalizing contracted military infrastructure, then nobody would make a contract with the military.
And while this may sound good to some, it sure wouldn’t be in Ukraine’s interest. Unreliable Starlink access is better than no Starlink access at all.
Ok, but “It would be great if people had to buy more of the thing” is not an accurate summary either. Putting a CD drive on a console does not mean you have to buy physical media.
[citation needed]
As I already said, fair use is generally not granted when it entails competition with the original work. See above regarding movie reviews vs copying an entire film.
It has nothing in common with it.
Legally, property is something that has an owner. IP has an owner, and like other types of property it can be transferred to another owner and become the subject of contracts.
IP cannot be “stolen”, and I never said it could be. Real estate cannot be “stolen” either, yet real estate is still property.
That’s all an AI needs in order to get trained on something. They just need to see it.
For someone who thinks other people are “weird” about legal language, you keep making the same mistakes.
When people “see” something, they do not need to create a copy of it in the legal sense. When I look at an old photograph, legally I do not create a copy of the photograph.
AI do not “just see” data. They need access to an electronic copy of the data. An AI cannot “see” an old photograph unless they first create a local copy of the photograph. There is a significant legal difference.
There are situations where permission is not required to use copyrighted material, mainly “fair use”. But AI training often does not qualify as fair use.
Otherwise, intellectual property is treated similarly to other types of property. For example, the person who owns a car can give you permission to use it. That doesn’t mean you can do whatever you want with it. The owner gets to set the rules. They aren’t “kings”, but as owners they do have near complete control over what you can do with their car.
When you upload something to social media, you (the content owner) give the host permission to display your content. That does not mean users who view your content have permission to do whatever they want with it.
There is plenty of open source code posted into repositories that are extensively mirrored, yet the code has lengthy conditions attached to it. If you use it in violation of the stated license, you could find yourself on the losing end of a lawsuit.
There are plenty of photographs posted onto Instagram, which is also designed to display them to anyone who asks. If a professional photographer finds that you’ve used one of their Instagram photos without permission, you could find yourself on the losing end of a lawsuit.
And the Fediverse may be a non-commercial decentralized platform, but copyright protection doesn’t magically disappear here. You give servers a license to display what you wrote, but you may reserve the same rights over your IP as software developers and photographers do over their own.
Copyright holders can prohibit all use of their work. If they can prohibit all use, then they can prohibit any specific use.
And they rarely give a wide-ranging “permission for their posts to be viewed by the public.” That’s like saying “I can do whatever I want with this source code, the developer gave permission for the code to be viewed by the public.” The legal language is often far more specific. There may be prohibitions on commercial usage, requirement for attribution, or even requirements that future code be licensed in a similar manner. There is almost always fine print when content appears to be “free”.
Of course, it’s possible that you could find a creative way around the legalese. Pointing a camera at a browser may work, until the fine print changes to prohibit that too. But anyway, that’s not what AI developers have done in the past. So one way or another, they may be forced to change their ways.
Hollywood and other big businesses will still find a way to train AI as usual. They are already used to paying musicians when a song is used in a movie. They can easily pay copyright holders for a license to use content as training data. It’s far safer - and more efficient - to pay up than try to get around the rules with a camera pointed at a screen. As a bonus, content creators who contribute training data may benefit from royalties.
Nevertheless, I think it will become more difficult for people who think they can easily get “free” training data from the web, just like 20 years ago when people thought they could easily get “free” music from Napster.
That’s a public performance, which is a form of redistribution. That’s not relevant to AI training.
Copyright law defines whether or not you can make a copy of a work. The person who owns the copyright can deny permission to make any copies, or grant you to make a permission to make a copy only under certain conditions. Those conditions are completely up to the copyright holder. They might prohibit public performance, but by no means is public performance the only thing that the copyright holder can prohibit. It’s simply a very common prohibition.
You are trying to trying to generalize from a specific right, viewing the content on a browser, to a general right to “look” at the content, to the right to train an AI. But legally those are not the same at all. You may be granted some, all, or none of those rights.
Suppose you are in a modern art gallery. You have been given the right to “look” at someone’s art. You can nevertheless be prohibited from making a photograph of the art, even if the camera is also “looking” at it. The owner of the art can attach whatever conditions they want to your photo, including how long you can keep it and exactly what you do with it.
For example you could be allowed to photograph the art for home use but not for wider distribution. You could be allowed to photograph the art for classroom use, but not for AI training. If you are not willing to follow all of the conditions, then you can’t make a photo of the art at all.
The same is true of text. Websites give permission to make a copy of their text for use on your browser. And they can set whatever rules they like for how else your copy may be used.
Except for being banned from using public data that non-American AIs are able to use.
Sure. Of course, America could also ban those non-American AIs from being used in the US. Just as America bans other products that infringe patents/IP.
Which is way more than what an AI model retains.
It makes no difference what the AI model retains. The only question is whether you had permission to use your copy in the manner that you did.
So for instance suppose you made a copy of a Disney movie in any fashion (by torrent, by videotaping a screening, by screen-capturing Disney+, etc), then showed it to a classroom in its entirety, and then deleted it immediately thereafter. You infringed copyright, because you did not have permission to use it in that manner even once. It makes no difference how long you retained your copy.
Note that it would also make no difference if there were actually no students in the classroom. Or if the students were actually robots. Or just one robot, or a software AI. Or if you didn’t use a screen to show the material, you simply sent the file electronically to the AI. Or if the AI deleted the file shortly after receiving it. You still didn’t have permission to use your copy in the manner you did, even once. Which means it was illegal.
America’s laws are not global laws.
True. But the GDPR has shown us that a country can take measures to protect its data globally.
If they wish to ban AI training this will become starkly apparent.
In every other field, researchers have long been required to use opt-in databases for their work. They can’t just “scrape” your medical records without your consent in order to study a particular disease. That would be wildly unethical.
Yet research, including AI research, has thrived in the US even with such ethical requirements. I am confident future AI researchers in America can be both ethical and successful.
Only very specific rights are needed, such as the right to learn concepts and styles from what you can see.
For AI training, you nearly always need a local copy of the data.
That’s the case for literally everything you view online. Putting it up on your screen requires copying it into your computer’s memory
Yes, and that copy is provided with restrictions. You can view your copy in a browser, but not necessarily use it for other purposes.
Every search engine ever has done this way more flagrantly than any AI trainer has. There have been plenty of lawsuits over this general concept already and it’s not a problem.
Those cases have delineated what Google is and is not allowed to do. It can only copy a short snippet of the page as a summary. This was ruled “fair use” largely because a short snippet does not compete against the original work. If anything it advertises the original work, just as movie reviews are allowed to copy short scenes from the movie they are reviewing.
On the other hand, AIs are designed to compete against the authors of the works they downloaded. If so, a fair use defense is unlikely to succeed.
Except that in this case it’s not torrenting a copy that Disney didn’t want to have online for you to see. It’s looking at stuff that you have deliberately put up online for people to see.
Disney does put its work online for people to see. So does the New York Times. That doesn’t mean you can make an unrestricted copy of what you see.
Besides, it’s actually not illegal to download a pirated movie.
Both are illegal in the US, although copyright holders generally prefer to go after uploaders.
Voice actors have a union.
Designers and engineers generally don’t. Yet.
Making something publicly available does not automatically give everyone unrestricted rights to it.
For example, you do not have permission to make copies of articles in the NYT even when they are available to the public. In fact, a main purpose of IP law is to define certain rights over a work even after it is seen by the public.
In the case of AI, if training requires making a local copy of a protected work then that may be copyright infringement even if the local copy is later deleted. It’s no different than torrenting a Disney movie and deleting your copy after you watched it.
Ok, so the main purpose of a blockchain is to get a bunch of computers that don’t trust each other to agree on who did what when.
A blockchain gives everyone a “voice”, so they can share what they heard. But it can’t be the sort of voice we have in the real world, because one computer can easily impersonate a million computers. In the case of Bitcoin, your “voice” is your computing power. Your computer might be able to impersonate a million computers, but your CPU cannot do the work of a million CPUs. So it is nearly impossible for single malicious computer to “drown out” all the other computers and insert a false message into the blockchain.
Bitcoin has a limit to the number of transactions per second because it wants computers to pause and talk to each other before validating a transaction. The delay is a feature, not a bug. This arbitrary limit is designed to self-adjust, so adding more computers won’t make the process go faster and removing computers won’t slow it down.
That said, computing power uses a lot of energy. Too much energy. So now there are other ways to assign voices that don’t rely on raw computing power but still prevent impersonation. This particular messaging protocol uses one of them, it is not based on computing power. But there is still a blockchain that contains everyone’s messages, and a malicious computer in theory cannot overwhelm the common consensus. So you can be pretty sure that if a message appears in the blockchain, then it was sent by the person who claims to have sent it.
Note that unlike Bitcoin, this blockchain is not public. It is not like a secret agent placing a classified ad that all can read (but not understand). Even the encrypted form of the blockchain can only be accessed by servers whose users participate in the conversation. If your server has nothing to do with a group, then you cannot glean anything at all about the conversations within the group.
You didn’t even know what a pgp key was before this convo or read receipts
Lol what? I knew what they were, I just thought it was stupid to bring them up because they solve nothing.
Falsifying evidence is a crime.
Oh, then there is no need to worry about it, I guess.
“why did you not respond”,
“Respond to what??”
they’re going to pull out their phone and say “see, I sent you this”.
Then the worker pulls out their phone and says, “see, it’s not on my phone”
That’s the contractors fault. Blockchain is irrelevant. If they didn’t check their email, they’re sure as hell not going to check a dumb ass blockchain.
Unless, of course, the sender/manager actually didn’t properly notify the contractor/employee, and now they are lying to cover their ass.
Like many disputes, it amounts to he-said-she-said. When it goes to court, the jury will flip a coin. There is a better way.
I just gave you one example of a use case. It’s hardly unique. There are plenty of time-sensitive messages sent in business settings, and plenty of people who don’t necessarily want to acknowledge receiving them.
More examples, off the top of my head:
Manager tells worker they need to cover an emergency on the weekend, worker claims they never received the message.
Business wants to cancel a work order, contractor shows up and says they weren’t properly notified of the cancellation.
Supervisor sends disciplinary note to employee before dismissal, employee says it was never sent and then claims wrongful termination.
And of course, this has nothing to do with email. So if you set up a “spam filter” that deletes your boss’s messages, that’s on you. They know they sent you the message, even if you delete it or otherwise pretend they didn’t.
This is about an independent audit trail, not “keeping logs”. Your personal email server doesn’t count, because you can alter the log to show whatever you want. Nobody is going to take your word for it.
Finally, it’s pretty clear you have no idea how this system is supposed to work, because you keep claiming that documents are “accessible by any third party”. You do understand that not every blockchain system is public, right?
Any form of internet communication is potentially susceptible to traffic analysis, so that flaw isn’t specific to this particular design.
The goals here are to address some of the other weaknesses of communication protocols, ie lack of auditability and reliance on a central server. They do not claim it’s completely impervious to attack.
Do you know what a read receipt is?
That doesn’t solve the problem. If you don’t get a read receipt, then you can’t prove you sent the message. And if the recipient doesn’t want you to be able to prove you sent a message, they can disable sending read receipts.
This sort of system is not meant for your use case. It is not meant for memes or other things nobody cares about. It is meant for people who need an auditable permanent copy of their communication.
For example, businesses sending orders, contracts, etc to each other. Or lawyers sending documents to each other. They need systems that are private, not susceptible to central server failure, yet nevertheless auditable in case of an untrustworthy recipient.
If a lawyer sends a time-sensitive letter to opposing counsel, the recipient must not be able to claim, “You did not send it to me on time”. Blockchain is a good solution for such needs.
I don’t want to leave my data publicly available with all the metadata
Did you read the paper? This isn’t Bitcoin. The metadata is not available to the public.
Even if it’s encrypted, it can be lost or destroyed if it’s stored client side.
I know what identity keys are, but they don’t solve the problem. If someone says they didn’t receive your message, the best way to prove you successfully sent it is to use a distributed ledger.
Storing client side isn’t good enough, your records could be lost or destroyed. That’s why people use Gmail.
And it’s not just third parties, what about untrusted recipients? For example, how do you prove you sent someone a message on a decentralized system?
But in this case, the metadata is not accessible to outsiders.
Yes, that sounds like their goal.
Blockchain is used to prevent a malicious participant from altering/corrupting records.
A key issue, often overlooked, is that US law imposes significant restrictions on the export and sale of military hardware.
Starlink is currently not considered military hardware. SpaceX is desperately trying to keep it that way, their ultimate goal is to sell subscriptions to civilians. Thus they get anxious when it is openly used for military purposes.
In this regard Starlink is somewhat similar to civilian GPS receivers, which automatically shut down at 1200 mph so they can’t be used in missiles.