It's not bending the knee, that's how they want it to be perceived, but what's really happening is that they're trying to pull up the ladder.
- What is OpenAI's level of copyright now?
- How is it going to be more "copyrighted" in the future?
- How does this affect competitors differently in the future vs. the copyright that OpenAI has now?
Limited. They’re hoping to change that. It’s no secret that open-source models are the long-run competition to the likes of OpenAI.
Nobody does. It’s being litigated.
They want it legislated. Model weights being proprietary by statute would close off the threat from “consumer-grade hardware” with “exaflop-scale.”
Then why did you say "Limited"? Surely YOU must have meant something by it when you said it. What did YOU mean?
I don't think you're saying that you are repeating something someone else said, and you didn't think they knew what they meant by it, and you also don't know what you/they meant. Correct me if I'm wrong, but I'm assuming you had/have a meaning in mind. If you were just repeating something someone else said who didn't know what they meant by it, then please correct me and let me know -- because that's what "nobody knows what it means" implies, but I feel like you knew what you meant so I'm failing to connect something here.
> It’s being litigated.
I'm not able to find any ongoing suits involving OpenAI asserting copyright over anything. Can you point me to one? I only see some where OpenAI is trying to weaken any existing copyright protections, to their benefit. I must be missing something.
I'm also unable to find any lobbyist / think-tank / press release talking points on establishing copyright protections for model weights.
Where did you see this ongoing litigation?
[1] https://www.bereskinparr.com/doc/chatgpt-ip-strategy
[2] https://hbr.org/2023/04/generative-ai-has-an-intellectual-pr...
From [1] > OpenAI’s Terms of Use, for example, assign all of its rights, title, and interest in the output to the user who provides the input, provided the user complies with the Terms of Use.
Re: [2]: I believe I referenced these specific concerns earlier where I said: "I only see some where OpenAI is trying to weaken any existing copyright protections, to their benefit. I must be missing something." This resource shows where OpenAI is trying to weaken copyright, not where they they are trying to strengthen it. It's somewhat of an antithesis to your earlier claims.
I notice you don't have a [0]-index, was there a third resource you were considering and deleted or are you just an avid Julia programmer?
Didn't say they do. I said "these are broad questions whose answers are worth serious legal time." I was suggesting one angle I would lobby for were that my job.
It's a live battlefield. Nobody is going to pay tens of thousands of dollars and then post it online, or put out for free what they can charge for.
> OpenAI’s Terms of Use, for example, assign all of its rights, title, and interest in the output to the user
Subject to restrictions, e.g. not using it to "develop models that compete with OpenAI" or "discover the source code or underlying components of models, algorithms, and systems of the Services" [1]. Within the context of open-source competition, those are huge openings.
> shows where OpenAI is trying to weaken copyright, not where they they are trying to strengthen it
It shows what intellectual property claims they and their competitors do and may assert. They're currently "limited" [2].
> notice you don't have a [0]-index
I'm using natural numbers in a natural language conversation with, presumably, a natural person. It's a style choice, nothing more.