So taking data without permission is bad, now?
Iām not here to say whether the R1 model is the product of distillation. What I can say is that itās a little rich for OpenAI to suddenly be so very publicly concerned about the sanctity of proprietary data.
The company is currently involved in several high-profile copyright infringement lawsuits, including one filed by The New York Times alleging that OpenAI and its partner Microsoft infringed its copyrights and that the companies provide the Timesā content to ChatGPT users āwithout The Timesās permission or authorization.ā Other authors and artists have suits working their way through the legal system as well.
Collectively, the contributions from copyrighted sources are significant enough that OpenAI has said it would be āimpossibleā to build its large-language models without them. The implication being that copyrighted material had already been used to build these models long before these publisher deals were ever struck.
The filing argues, among other things, that AI model training isnāt copyright infringement because it āis in service of a non-exploitive purpose: to extract information from the works and put that information to use, thereby āexpand[ing] [the worksā] utility.āā
This kind of hypocrisy makes it difficult for me to muster much sympathy for an AI industry that has treated the swiping of other humansā work as a completely legal and necessary sacrifice, a victimless crime that provides benefits that are so significant and self-evident that itās wasnāt even worth having a conversation about it beforehand.
A last bit of irony in the Andreessen Horowitz comment: Thereās some handwringing about the impact of a copyright infringement ruling on competition. Having to license copyrighted works at scale āwould inure to the benefit of the largest tech companiesāthose with the deepest pockets and the greatest incentive to keep AI models closed off to competition.ā
āA multi-billion-dollar company might be able to afford to license copyrighted training data, but smaller, more agile startups will be shut out of the development race entirely,ā the comment continues. āThe result will be far less competition, far less innovation, and very likely the loss of the United Statesā position as the leader in global AI development.ā
Some of the industryās agita about DeepSeek is probably wrapped up in the last bit of that statementāthat a Chinese company has apparently beaten an American company to the punch on something. Andreessen himself referred to DeepSeekās model as a āSputnik momentā for the AI business, implying that US companies need to catch up or risk being left behind. But regardless of geography, it feels an awful lot like OpenAI wants to benefit from unlimited access to othersā work while also restricting similar access to its own work.
Your mistake is thinking that ādataā and ācopyrightā or āownshipā are the same thing. They arenāt
You can download a song, and thus be in possession of the data of that song, and you can even copy the file within the parameters of copyright law.
However, simply having the data is not the same thing as owning or holding a license to the song itself, and so you are in violation of the law (where I live, at least) if you try to distribute that song or use it in a non-fair-use context.
IF you were to copy my work and exploit it in a for-profit context for millions of dollars (and you happened to be operating in a region in which applicable copyright laws happen to apply) youāre damn right I would come after you for a slice of the pie, and I would almost certainly win. Just copying what I say and pasting it in a quote isnāt something that I can prove damages on, because it isnāt something youāre profiting on in any way, so the idea of āenforcingā it is irrelevant and obviously not worth it.
This is where we are going to have to disagree. I am absolutely willing to fight fire with fire by using the copyright system against big tech. I donāt make the rules, but IF rules are to exist in terms of what is or is not fair use of copyrighted material, then I DO expect those rules to apply equitably. (Whether they will or not remains to be seen, but letās see what precedent gets set and Iāll adapt from there.)
Can I ask you a personal question: what do you create, and do you submit it to the public domain?
As for me, I write music, create art, make games and write computer code and do a number of other things that I absolutely claim ownership over. So, when I write a song or paint a picture who the fuck is anyone else to try to take that away from me or claim it as something that they own and control? Iāve written thousands of lines of GPL code and contributed to many hippy-dippy open source free software projects over my lifetime, and even in that kind of copyleft context we still maintain a copyright over the code we right (as seen at the top of every source and header file).
I only ask because I find that the people who are most pro-AI and most anti-copyright are generally people who have never created anything of their ownātheyāve written no songs, theyāve drawn no pictures, theyāve written no storiesāand now they incorrectly generative see AI as something that āevens the playing fieldā by compensating for their lack of skills and drive.
But Iāll repeat myself, AI isnāt ushering us into a post-copyright world where the little guy is empowered in anyway. Itās just a punch of useful idiots downloading completely proprietary binary blobs from the biggest, richest corporations, fooling themselves into thinking that theyāre being empowered to create things when in reality theyāre just beta testing a plagiarism machine on a industrial scale thatās designed to enrich the richest.
Iām a software engineer and I have been playing guitar nearly every day since I was 8 years old. I release everything GPL/AGPL or CC-BY-SA that I own and can. Heck, I am racking every day trying to figure out ideas that can hopefully make me a living while also giving everything I have away. I donāt want to own my shit man, I just want to share what I have and hope itās useful, and I donāt want people being assholes so I opt for the copyleft instead of liberal licenses.
Even in GPL and CC-BY-SA context you still retain copyright ownership over your work. I write GPL code for a modest living, and my real name copyright goes on everything I write. Likewise, your still asking to be credited in your CC-BY-SA music. Nothing wrong with that.
The point being is that we are making a conscious decision to license the things we create in a permissive way. Neither of us are anonymously dumping our work into the public domain because clearly we do care about ownership and copyright. Thatās well within our rights as creators.
Generative AI is exploiting our work and not even doing the bare minimum of following the licenses that we shared them under.
I only care about copyright while it exists. If copyright is abolished and thereās no such thing as intellectual property then Iām happy with that as well.
No one should have the right of using police with guns to maintain the ownership of ideas and our culture. I donāt want this power. I donāt want you to have this power over my neighbors. I donāt want the Disney corporation having this power over my friends and family. But if capitalists wield that power against the common man, then Iām not against the common man wielding it against the capitalist.
Iām okay with using copyright against only those that have more power than me. I donāt ever want to threaten a normal human being with it, only capitalists.
Iām not against all cases of generative ai, code or visual. Iāve had legit use cases for them, and in a post scarce world we wouldnāt care about these things being made in a completely floss way. If copyright were to last only 15 years after publication then I think the world would be much better and we wouldnāt be having this conversation. But I wonāt argue ai stuff as it currently standsāit is a grift or a stockpiling.
I just donāt like the premise of a market where one has to sell their artistic labor in order to survive, or thrive. Iām on board with noncommercial licenses and everything because the reality looks different, but that was not my point. And neither was it the point of the original comment you replied to.