The US ruled that AI produced things are by themself not copyrightable.
So no, it doesn't belong to OpenAI.
You might be able to sue for penalties for breach of contract of the TOS, but that doesn't give them the right to the model. And even if it doesn't give them any right to invalidate unbound copyright grants they have given to 3rd parties (here literally everyone). Nor does it prevent anyone from training their own new models based on it or prevent anyone from using it. Oh, and the one breaching the TOS might not even have been the company behind DeepSeek but some in-between 3rd party.
Naturally this is under a few assumptions:
- the US consistently applies it's own law, but they have a long history of not doing so
- the US doesn't abuse their power to force their economical opinions (ban DeepSeek) on other countries
- it actually was trained on OpenAI, but uh, OpenAI has IMHO shown over the years very clearly that they can't be trusted and they are fully in-transparent. How do we trust their claim? How do we trust them to not retrospectively have tweaked their model to make it look as if DeepSeek copied it?
The copyright office ruled AI output is uncopyrightable without sufficient human contribution to expression.
Prompts, they said, were unlikely enough to satisfy the requirement of a human controlling the expressive elements thus most AI output today is probably not copyrightable.
>The Office concludes that, given current generally available technology, prompts alone
do not provide sufficient human control to make users of an AI system the authors of the
output.
Prompts alone.
But there are almost no cases of "Prompts Alone" products seeking copyright.
Even what 3-4 years ago?, AI tools moved into a collaborative footing. Novel AI forces a collaborative process (and gives you output that can demonstrate your input which is nice). ChatGPT effectively forces it due to limited memory.
There was a case, posted here to ycombinator, where a chinese judge upheld "significant" human interaction was involved when a user made 20-odd adjustments to their prompt iterating over produced images and then added a watermark to the result. I would be very surprised if most sensible jurisdictions didn't follow suit.
Midjourney and ChatGPT already include tools to mask and identify parts of the image to be regenerated. And multiple image generators allow dumb stuff like stick figures and so forth to stand in as part of an uploaded image prompt.
And then theres AI voice which is another whole bag of tricks.
>thus most AI output today is probably not copyrightable.
Unless it was worked on even slightly as above. In fact it would be hard to imagine much AI work that isn't copyrightable. Maybe those facebook pages that just prompt "Cyberpunk Girl" and spit out endless variations. But I doubt copyright is at the forefront of their mind.
A person collaborating on output almost certainly still would still not qualify as substantive contributions to expression in the US.
The US copyright's determination was based on the simple analogy of someone hiring someone else to create a work for them. The person hiring, even if they offer suggestions and veto results, is not contributing enough to the expression and therefore has no right to claim copyright themselves.
If you stand behind a painter and tell them what to do, you don't have any claim to copyright as the painter is still the author of the expression, not you. You must have a hand in the physical expression by painting yourself.
>A person collaborating on output almost certainly still would still not qualify as substantive contributions
But then
>You must have a hand in the physical expression by painting yourself.
You contradict yourself. Novel AI will literally highlight your contributions separately to the AI so you can prove you also painted. Image generators literally let you paint over the top to select AI boundaries.
but even then wouldn't the people using OpenAI still be the author/copyright holder and never OpenAI? (as no human on OpenAIs side is involved in the process of creating the works)
OpenAI is a company of humans, the product is ChatGPT. Theres a grey area regarding who owns the content, so OpenAI's terms and conditions state that all ownership of the resulting content belongs to the user. This is actually advantageous because it means that they dont hold ownership on bad things created by their tool.
That said you can still provide terms to access the tool, IIRC midjourney allows creators to own their content but also forces them to license it back to midjourney for advertising. Prompts too from memory.
So no, it doesn't belong to OpenAI.
You might be able to sue for penalties for breach of contract of the TOS, but that doesn't give them the right to the model. And even if it doesn't give them any right to invalidate unbound copyright grants they have given to 3rd parties (here literally everyone). Nor does it prevent anyone from training their own new models based on it or prevent anyone from using it. Oh, and the one breaching the TOS might not even have been the company behind DeepSeek but some in-between 3rd party.
Naturally this is under a few assumptions:
- the US consistently applies it's own law, but they have a long history of not doing so
- the US doesn't abuse their power to force their economical opinions (ban DeepSeek) on other countries
- it actually was trained on OpenAI, but uh, OpenAI has IMHO shown over the years very clearly that they can't be trusted and they are fully in-transparent. How do we trust their claim? How do we trust them to not retrospectively have tweaked their model to make it look as if DeepSeek copied it?