5 ms·
Paraphrasing is not the issue. The issue is that OpenAI copied the Times’ creative works into a GPU to train a model. That copy was likely neither licensed nor
by twoodfin 3y ago
Paraphrasing is not the issue. The issue is that OpenAI copied the Times’ creative works into a GPU to train a model. That copy was likely neither licensed nor fair use.
- dragonwriter 3y agoPersonally, I think the Times has a far better case for presenting mechanical copies (including with mechanical alteration) than it does with model training. The tool building of model training is more likely to be fair use than the use of the tool to provide mechanical copies of copyirght-protected material that competes directly with the original in the market.
- twoodfin 3y agoWhat makes you think that? The most relevant case is probably MAI Systems, where even a copy of a licensed program into RAM for an unlicensed use (diagnosis and repair by a third party) was deemed a violation. Congress took the ruling seriously enough to carve out a fair use exception for that use. I don’t think there’s any comparable exception here, especially when the ultimate purpose is to build a commercial product capable of producing works (news reports) in precisely the same market as the original works.
- swalling 3y agoFair use only covers very limited circumstances which probably does not include selling a subscription (ChatGPT+). If you’re selling a repackaged reproduction of someone else’s copyrighted works, that’s never protected by fair use.
- Scion9066 3y agoBut you generally can't copyright the underlying facts, only the creative expression in the article/work itself. If they can get it to extract just the facts and how they relate to one another, that wouldn't really be protected by copyright. They might try to bring back the "Hot News" doctrine though.
- swalling 3y agoThe models not only were trained on copyrighted works, but when prompted will substantially and repeatedly reproduce copies of creative work (not just facts). That's infringement. https://www.theverge.com/2023/7/9/23788741/sarah-silverman-openai-meta-chatgpt-llama-copyright-infringement-chatbots-artificial-intelligence-ai https://www.theverge.com/2023/7/9/23788741/sarah-silverman-o...
- nomel 3y agoMy naive reading of it sees that it only applies for time-sensitive facts. Considering the training times required, and knowledge cutoffs, could this apply?
- bandrami 3y agoBut you can't copyright the underlying facts nor the style of writing, just the literal words used in your article. If you can't point to a sequence of words in the original that's in the later work, that later work isn't "derivative".
- asdfman123 3y agoThe question though is if we want to stifle innovation by requiring LLMs get permission from every single relevant party on the internet. And I usually lean anti-corporate too, but banning people in the US from using data for LLMs might just mean they start being trained somewhere else that doesn't care as much about US law.
- Tao3300 3y agoIf I read it and memorize it, my brain has made a copy.
- twoodfin 3y agoNot as far as copyright law is concerned. If you then use your brain to write it back down—or sing it as a song in Central Park—now you have created a copy under the law.
- Tao3300 3y agoRight. But the comment above said the issue is with the copy made for training. The "read it and memorized it" copy, not the "write it back down" copy.
- twoodfin 3y agoThat copy didn’t go into a human’s brain. It went into GPU memory. It’s a copy under the law, no different from copying a Taylor Swift mp3 onto a flash drive. Whether that copy was fair use is the key question.
- prepend 3y agoIf I buy a license to listen to a Taylor Swift mp3, I can copy it onto a flash drive (or anywhere I like to use it). And that’s fine until I distribute copies to others.
- twoodfin 3y agoYes, and if OpenAI had a license to use the Times’ works as model training material this would be a non-issue, of course.
- bandrami 3y agoThere's really not a "license" in that sense for text. There's not some magical way you can use copyright to protect a probability model of the words you make, because that's style and style is not copyrightable (nor are the underlying facts the NYT reports).
- CuriouslyC 3y agoDid the Times grant a license to every router on the internet to transmit its intellectual property to other routers? If not, the judge should grant an injunction contingent on requiring the Times to verify that every person who accesses their content is doing so only over routers and other devices with express written authorization, for every step in the process. Maybe even extend it to browsers and client libraries for encoding/decoding, to be accurate. I say if people want to come up with bullshit lawsuits, we should play into them and force those people to suffer the consequences of said bullshit.
- twoodfin 3y agoYou don’t need a fair use exemption for transient copies in service of licensed or fair uses. Computers and networks have been around a long time. These issues have been given a good workout.
- mike_hearn 3y agoBut the copy used during training is itself transient, so all this boils down really to the question of whether training a machine is fair use. Which can't be answered here exactly because the concept of fair use is deliberately vague, so this will boil down to a lawsuit and probably go to the Supremes. The USA will work something out that's reasonable as they always do and, lacking AI companies and often the concept of fair use to begin with, the rest of the world will never work out the necessary case law and fall even further behind. Possible exception: UK, which does have some significant LLM companies and also fair use law, albeit as is often the case with the UK these firms are US/UK hybrids with significant presence in both countries and the legal HQ is usually in the USA.
- bandrami 3y agoIt's not even a question of "fair use" because OpenAI isn't providing anybody a copy. Probability models really just aren't copyrightable to begin with.
- 3y ago
- prepend 3y agoDo you think OpenAI doesn’t subscribe? I too load creative works into all sorts of temporary structures in order to read the paper. I don’t need to license it, I pay for a subscription. Should I pay more if I memorize the paper? Should I pay more if I read it to my sick friend in the hospital? Should I pay more if I save copies to my own hard drive and grep for words in the files? Should robots have a higher subscription price? This whole “they copied it into a gpu” doesn’t matter. People read and interpret. Robots read and interpret. I don’t want to live in a world where every specific device and use needs to be licensed. Especially ex post facto. That will suck so hard.
- version_five 3y agoShould I pay more if I memorize the paper? Should I pay more if I read it to my sick friend in the hospital? It's not even close to either of those things. It's "should I pay more if it infinitesimally affects my perception of english grammar or knowledge of a subject". The llm isn't, for any functional purpose, memorizing, it's getting weight updates as it learns from these examples which are teaspoons of information in a sea of trillions of tokens.
- twoodfin 3y agoHow about “they copied it into a terrific text-to-speech engine and sold ads against the resulting podcasts”?
- prepend 3y agoReading a text has been ruled as performance and a copyright violation. I think the issue is that LLMs don’t make a copy or distribute a copy. They use the content to create something else. I don’t remember the copyright term for whether is is transformative enough. But it basically says I can’t copy Starry Night, but I can create a painting with the same color scheme and themes as long as it’s different enough from the original.
- rurp 3y agoAre you using this knowledge to produce a product that materially undercuts NYT revenue?
- kolinko 3y agoGoogle does the same to produce a search index.
- twelve40 3y agoyou can easily opt out of that, or control it to your heart's desire (including what snippets to show), and it will be honored. There is no way to opt out of this bullshit. So no, not the same at all.
- bandrami 3y agoThat's a courtesy of the search engine, not a requirement of copyright.