> Disney couldn't then claim that Claude does not in fact have permission because Disney's use of the tool in such a way mean Disney now has unclean hands when bringing the claim (or atleast Anthropic would be able to use it as a defence).
Disney wouldn't be able to claim copyright infringement for that specific act, but it would have compelling evidence that Claude is cavalier about generating copyright-infringing responses. That would support further investigation and discovery into how often Claude is being 'fooled' by other users' pinky-swears.
> Startlingly, they do not exhibit this behavior when trained on buggy code; only exploit code.
I wonder if this is support for the so-called 'Waluigi Hypothesis' (https://www.alignmentforum.org/posts/D7PumeYTDPfBTp3i7/the-w...). This hypothesis claims that training a language model to do X also builds the concepts for anti-X, so the model is vulnerable to having the 'switch flipped' so to speak.
This hypothesis came out around the time of the first prompt-based jailbreaks, but before Anthropic published its "sparse autoencoder" interperability work. Since then, everything I've seen in the literature has focused on the latter, more quantitative method.
I read the Waluigi proposal and played around with the concepts at the time. It seemed effective. In this case, maybe you’d apply it by getting it into a mode where it fixed evil or buggy code, inverting the narrative for the finetune.
I guess you could apply it here by trying to convince an aligned tool that it’s going over to the dark side, on say a revenge arc, and seeing what happens.
> The RIAA lawyers never had to demonstrate that copying a DVD cratered the sales of their clients. They just got high penalties for infringers almost by default.
The argument for 'fair use' in DVD copying/sharing is much weaker since the thing being shared in that case is a verbatim, digital copy of the work. 'Format shifting' is a tenuous argument, and it's pretty easily limited to making (and not distributing) personal copies of media.
For AI training, a central argument is that training is transformative. An LLM isn't intended to produce verbatim copies of trained-upon works, and the problem of hallucination means an LLM would be unreliable at doing so even if instructed to. That transformation could support the idea of fair use, even though copies of the data are made (internally) during the training process and the model's weights are in some sense a work 'derived' from the training data.
If you analogize to human leaning, then there's clearly no copyright infringement in a human learning from someone's work and creating their own output, even if it "copies" an artist's style or draws inspiration from someone's plot-line. However, it feels unseemly for a computer program to do this kind of thing at scale, and the commercial impact can be significantly greater.
You can't make valid legal analogies to human learning when dealing with copyright law, because human brains are not a fixed media under copyright law, thus impressions in human brains are not copies of any kind under copyright law, thus "well, when you make an impression in the human brain, its not a copyright violation" is never a good legal analogy for when you make something that is in a form which can be a copy under copyright law.
> there's clearly no copyright infringement in a human learning from someone's work and creating their own output, even if it "copies" an artist's style or draws inspiration from someone's plot-line.
What do you mean here by "clearly?" This is not at all clear, and court cases have been decided in the opposite direction.
In the case you linked, it wasn't infringement for Robin Thicke to be impressed by Marvin Gaye. It was infringement to create a substantial copy.
Applying this same legal doctrine to LLMs, it's totally fine for them to train on copyrighted works. A problem would only arise if they produced substantial copies of those works.
It may be difficult to quantify the harm done in that case. Robin Thicke sold millions of copies of blurred lines. Each LLM output is generally consumed by a single person. There could be millions of examples of blatant copyright infringement by LLMs that will never come to light in court. There also could be vanishingly few. We may need new regulations to police it.
But that case involves publishing of content. If you want to compare them, it seems like you'd argue that the content of Pharrell Williams' head is somehow copyright infringement. I've yet to see anyone credible say that an AI model outputting exact content and then someone publishing that should be allowed. If you manage to make an AI model output copyrighted content, you can't then claim it's not copyrighted. If you sit on one side of a table and read a book out loud and I write it down (with minor transcription errors), that content is almost certainly still copyrighted content and I could not distribute it legally.
It matters because that is what the plaintiffs are arguing - that the use of copyright material for AI training is not fair use. They are not solely arguing that meta engaged in copyright infringement through piracy.
Information contained in subjective mental experience is not a medium which can qualify as a copy (infringing or otherwise) in copyright law, whereas data recorded in digital media such as computer memory is, so they are not similar circumstances with regard to copyright law, however analogous you might feel they are from some other perspective.
> Thomson Reuters has won the first major AI copyright case in the United States. In 2020, the media and technology conglomerate filed an unprecedented AI copyright lawsuit against the legal AI startup Ross Intelligence. In the complaint, Thomson Reuters claimed the AI firm reproduced materials from its legal research firm Westlaw. Today, a judge ruled in Thomson Reuters’ favor, finding that the company’s copyright was indeed infringed by Ross Intelligence’s actions.
Though a cynical way to read this would be that the larger corporation won against the smaller, so the actual precedent here is Meta winning.
But as others have noted in these comments, even that has a silver lining: should Meta win, then everyone can pirate books to read, right? /s
If I remember right, that particular case was Ross Intelligence almost copying a database of Westlaw's summaries completely and with the intent to compete directly too, with their only real addition being adding AI to the search capabilities, making it much more nuanced of a decision than just ruling against the use of AI.
> There are things to sort out, but it's certainly doable. They can probably use direct COGS to determine the tariff basis.
You could construct a variety of accounting methods to calculate a tax on the 'foreign content' of movies, but how do you actually impose this tax?
A tariff on physical goods is easy. A country requires that goods enter the country through customs facilities, and then the nice customs official doesn't release the thing until the tax is paid. The legal and physical ability to impose these taxes is long, long established.
How do you physically impose a tariff on a movie? If the master is transported physically, what is its value? The value of the fixed copy/master doesn't necessarily include the value of the IP involved, in the same way that a DVD of a $500 million movie might have a retail value of $20. What about movies transferred digitally, since there are no customs checkpoints on fiber-optic lines?
What legal apparatus would be used to impose this tax? Trump is currently getting away with the physical-goods tariffs because the legal infrastructure to collect the taxes is already in place, and remaining legal disputes are just about whether the President can unilaterally set or change tariff rates. If you'd need new law to "tariff movies," then the chance of this whim turning into a real tax drops sharply.
> Surely there are people trading in these contracts that... don't want their oil delivered to Cushing?
A big-enough buyer will know how to get oil from Cushing to their facility, often by pipeline. One who doesn't really want oil in Cushing is likely to close out their futures trade before the settlement date, treating it like a purely financial transaction.
> It seems like such a bizarre setup to literally require all the oil to come to this one specific town, I assume I'm missing something obvious?
Futures contracts need to be based on the price of something, but the price of a physical good depends on location. Delivery of a barrel of crude to the South Pole would be much, much more expensive – and more variable – than delivery to a big oil terminal. Contracts for physical goods need some kind of agreed-upon reference point, even if most of the time things get financially settled without delivery.
> All futures should be settled by the actual commodity!
Why? The legitimate hedging role of futures and options is often financial in nature, even for physically-settled contracts.
Take West Texas Intermediate as an example. That's a physically-settled contract, with delivery in Cushing, Oklahoma.
What if I want to lock in a future price of oil but I'm not in Cushing, Oklahoma? Nobody's going to create a liquid futures market with delivery to my loading dock, but most of the time I can get oil on the spot market from a local supplier that already includes/amortizes the transportation cost.
It's far better for me to use the liquid futures market for hedging and still buy on the spot market, closing out the futures contract before delivery. For me, it's as if the futures market is cash-settled, even with a completely non-speculative transaction.
> You know a burger tastes really good when you eat it, no benchmarks required.
I'd say this is a good example of the opposite, where the problem is finding the quantification of an ultimately subjective experience. Take three restaurant reviewers to a burger joint and you might end up with four different opinions.
Benchmarks proliferate because many LLM domains defy easy, quantitative measurement, yet LLM development and deployment are so expensive that they need to be guided by independent and quantitative (even if not fully objective) measures.
> Why should I be bothered to read what nobody could be bothered to write?
In an art-criticism sense I broadly agree with you, but I think you go too far from a reader-experience point of view.
> The point of writing is communication and creative writing is specifically about the human experience
That's a point of writing, but the point of reading is only sometimes about communication. It's also about entertainment, enrichment, or expressing half-formed thoughts and feelings.
Take a step back from the autonomously-written novel and imagine something a bit more collaborative. Many players of open-world-ish games develop an emergent story, what if that story could be semi-automatically novelized to document the unique narrative of the playthrough?
SimCity 2000 had "mad-lib" newspaper articles that commented on the city's status; consider ones written by an LLM with full knowledge of the game's state and context.
> something a LLM can mimic but never speak to authoritatively.
Supposing a reader can't tell the difference between an average human-written novel and an average LLM-written novel, where does this authority lie?
> Reciprocal tariffs are calculated as the tariff rate necessary to balance bilateral trade deficits between the U.S. and each of our trading partners. This calculation assumes that persistent trade deficits are due to a combination of tariff and non-tariff factors that prevent trade from balancing.
You know, I need to put a tariff on my local grocery store. They keep taking advantage of me by selling me food, when I export nothing to them!
Half of US imports from Madagascar is vanilla. It is the only place in the world that grows a lot of good vanilla. All you will be getting is more expensive vanilla.
> You make money on your laptop which you then trade for a variety of goods from actual producers
Isn't that actually also a good analogy for the way the US produces a lot of money "on its laptop" (service industry) and imports "actual" goods like consumer items from overseas?
> I started wondering if there's a way for citizens to push the issue that these tarrifs are unconstitutional.
Anyone who imports and is assessed a tariff would have personal standing to sue, arguing that the tax is illegal. I think there are some procedural requirements first to e.g. formally complain, but the dispute would eventually get its day in court.
However, these tariffs are being applied at least under a thin veneer of law regarding declared emergencies. It's clearly beyond the intent of the law for Trump to do this, but it's hard to draw a bright line between this and (e.g.) ambitious environmental regulation.
Disney wouldn't be able to claim copyright infringement for that specific act, but it would have compelling evidence that Claude is cavalier about generating copyright-infringing responses. That would support further investigation and discovery into how often Claude is being 'fooled' by other users' pinky-swears.
reply