Re:NY Times copyright suit wants OpenAI to delete all GPT in...
by 1vuio0pswjnm7 on Thursday December 28, 2023 @10:39PM
(#38801551)
If there is payment then usually there is an agreement. An agreement can limit fair use. Can the NYT, via an agreement, e.g., "Terms of Use", limit what the subscriber does with the articles. There is not much precedent that suggests otherwise.
Consider the analogy from libraries that want to do data mining.
"Unfortunately, in licenses for digital scholarly content the majority of content acquired by research libraries publishers often include terms that prohibit certain uses that would otherwise be allowable under the Copyright Act. For instance, licenses may require libraries or individual researchers to negotiate for otherwise lawful activities, such as text and data mining, and to pay exorbitant fees on top of the cost of the content itself. While new regulations allow researchers to circumvent technological protection measures to access copyrighted materials, licenses for that content may include terms that explicitly prohibit this circumvention. In many cases, these activities might actually increase the value of published material; for instance, if a data-mining project yields new knowledge about a topic covered in a journal, it may very well spark new interest in that journals content. Libraries and publishers have often assumed that license terms that restrict copyright exceptions are enforceable under state contract law. There is, however, surprisingly little case law on this point."
Putting some string in a robots.txt to try to stop data collection is an amusing "solution". Should copyright owners have "Terms of Use" that limit usage for commercial "AI" purposes.
Consider the analogy from libraries that want to do data mining.
"Unfortunately, in licenses for digital scholarly content the majority of content acquired by research libraries publishers often include terms that prohibit certain uses that would otherwise be allowable under the Copyright Act. For instance, licenses may require libraries or individual researchers to negotiate for otherwise lawful activities, such as text and data mining, and to pay exorbitant fees on top of the cost of the content itself. While new regulations allow researchers to circumvent technological protection measures to access copyrighted materials, licenses for that content may include terms that explicitly prohibit this circumvention. In many cases, these activities might actually increase the value of published material; for instance, if a data-mining project yields new knowledge about a topic covered in a journal, it may very well spark new interest in that journals content. Libraries and publishers have often assumed that license terms that restrict copyright exceptions are enforceable under state contract law. There is, however, surprisingly little case law on this point."
https://www.arl.org/wp-content/uploads/2022/07/Copyright-and...
Putting some string in a robots.txt to try to stop data collection is an amusing "solution". Should copyright owners have "Terms of Use" that limit usage for commercial "AI" purposes.