Hacker News Reader

Comment by sva_ | Hacker News Reader

I wonder if when you train a model based on the weights of another model, do the owners of the original model have any sort of copyright claim? They trained it on data on which they had no copyright claim either, fwiw.

View on HN

Why bother with the weights when you can just take the literal output of another model and consume as part of your training corpus?

What kind of copyright claims can be made against the content generated by a model wielding the desired weights?

In fine-tuned models, for example, you may only train a real subset of the weights. Who is the copyright owner now and what is the justification?

There might be other ways of transfer-learning where you can greatly benefit from the pretrained weights, some of which may have not yet been discovered.

> In fine-tuned models, for example, you may only train a real subset of the weights.

I am not sure to what degree this applies throughout the various offerings. We are finding that "fine" tuning cuts pretty damn deep. For instance, with OpenAI you can start with base Curie/Davinci and build an effective chat bot that makes ChatGPT look incompetent in a very narrow vertical slice of some specialized domain.

According to the US Copyright office's recent ruling, nothing generated by these models can be copyrighted (!), so in theory you should be fine!

nit: that wasn't a "ruling" so much as it was "guidance" [0] [1]

[0] https://www.federalregister.gov/documents/2023/03/16/2023-05...

[1] Discussed on HN: https://news.ycombinator.com/item?id=35191206

The position of the US copyright office is based in large part on prior case law (from your first link).

Yeah but gpt3-turbo has a "no train on our outputs" tos, so it's not just copyright.

Let's say company A uses OpenAI models to create an open dataset, like alpaca. If company B uses that dataset to fine tune a model, is company B liable to TOS violation? Afterall, they may not even be an OpenAI customer.

Does it actually? The only thing I ever saw was "don't use our models to build competing models". Which is clearly ambiguous at best when dealing with generative natural text. This also implies to me that using their models to help with fine tunes of their other offerings would be ok.

They're the first word on this, not the last, so it doesn't mean much. It also clearly leads to absurd results for text models, which can memorize their training data, or echo back your inputs. Or you could pass their output through a human to copyright it.

Not all countries have the same view on this either. A few countries, including the UK, maintain the view that the AI's operator, owner or custodian hold the copyright instead.

As I understand it, it's not a copyright issue, it's a terms-of-service issue. You're only authorized to use GPT in accordance with their terms, which prohibit using the service to develop a competitor. Doing so anyway is unauthorized use of the service, which is a Computer Fraud and Abuse Act violation, independent of what is and isn't copyrightable.

This is an Americanism. In many parts is Europe copyright claims over copyright from the US are honoured, but yet, violation of TOS by itself does not put you in risk of criminal lawsuit.

There are also plenty of places around the world where US copyright is not honored.

This includes, arguably, the US itself, though copyright violation is a lot more blatant in some parts of the world, which don't even offer the fig leaf of trying to enforce US copyright in their legal systems.

Technically only three:

https://legalbeagle.com/6780679-countries-copyright-laws.htm...

One, the Republic of San Marino was founded in 301 AD and thus has its own laws that predate the US copyright laws by some 1400+ years.

Is there any reason their near 2,000 year old freedoms should be constrained by the whims of a teenage republic?

Just make your millions and fight in court for a hundred years whilst they judges spend years learning ML.

Are google and oracle are still fighting over java?

There's probably not a lot of hope for the litigation on hyper niche bits of config.

Reply on news.ycombinator.com