Copyright: The New York Times attacks ChatGPT while OpenAI positions itself as defender of the press
December 27, 2023: The New York Times, a renowned American newspaper, files a complaint regarding the use of millions of articles reserved for its subscribers by the conversational agent ChatGPT. OpenAI, the creator of this artificial intelligence, issued a response on its website on January 8, 2024. Here's what their teams state:
ChatGPT, Supporter of the Press
OpenAI indicates that it has extensively collaborated with press organizations to design and improve its conversational robot. According to the firm, such a tool serves both its own interests and those of the press. It would automate the most time-consuming tasks for journalists, prioritize real-time news, and offer media new ways to communicate with their readers.
Such collaboration also helps feed the algorithm with articles reserved for subscribers to better understand the current context. This is precisely what led the New York Times to file a complaint for violation of its copyright.
A Tool Compliant with Regulations
Regarding the appropriation of public press articles, the company argues that the training of its algorithms adheres to fair use.
In the United States, fair use is a set of legal rules, originating from legislative and jurisprudential sources, that provide limitations and exceptions to the exclusive rights of the author over their work (copyright). It seeks to balance the interests of copyright beneficiaries and the public interest in the distribution of creative works, authorizing certain uses that would otherwise be considered illegal.
Source: Wikipedia
OpenAI adds that many countries, as well as the EU, have enacted laws allowing artificial intelligences to train on original works.
Aside from the legal aspect, OpenAI advocates the duty to provide access to information for millions of people. However, the tool allows newspapers to exclude links from their sites in the ChatGPT training model.
The New York Times Allegedly Manipulated ChatGPT to Induce "Regurgitations"
It happens that ChatGPT reproduces certain texts word for word in its responses, a phenomenon called "regurgitation". OpenAI describes it as a rare bug and specifies that the tool learns from making mistakes, while its developers ensure that it does not happen again.
Although the New York Times directly accused the tool of plagiarism, OpenAI raises that no specific article has been reported as evidence. According to the company, the newspaper would have used specific prompts multiple times to push the tool to plagiarism.
The legal battle that is unfolding, in the field of copyright, could have major repercussions in the coming years. Both for press organizations and for companies creating and developing artificial intelligence.
References:
- France 24 - ChatGPT et droits d’auteur : le New York Times, héraut de la liberté de la presse
- OpenAI - OpenAI and journalism
- OpenAI - Procédure de retrait
- Directive (UE) 2019/790 du Parlement Européen et du Conseil du 17 avril 2019 sur le droit d'auteur et les droits voisins dans le marché unique numérique
- Fair Use - Wikipedia
[Cover Photo: Natalia Y.]
Support us by sharing the article: