The New York Times Office Building./Unsplash
Gen AI has been pretty hacked since its launch. Since it is an innovative product, there is no relevant legal ruling yet other than that which states that copyright cannot be taken or attributed to works created by AI, including images, photographs, music videos, songs, films, etc.
The prevailing assessment is that most of the Western world, including Israel, has aligned itself with the spirit of American case law according to which copyright cannot be claimed for the creation of a machine, but when it comes to the machine itself, today the perception that the rights to AI creations belong to the machine has become quite established.
However, something about the fact that AI takes existing content and creates new ones from it has created explosive potential for lawsuits. Last week, the New York Times filed a lawsuit against Open AI and its partner Microsoft for copyright infringement.
The lawsuit deals with the "study" phase of the machine, during which the machine absorbs as much data as possible, in order to base its products on it. The wider the reservoir, the better and higher quality the machine will be. According to The New York Times, which is known to be one of the largest and most prestigious newspapers in the world, the AI tools of both companies were trained on original data created by the newspaper's reporters and therefore infringe the paper's copyright. Are there things in Go?
Journalism and artificial intelligence - our relationship to where?
There are some interesting aspects of the relationship between the journalism industry and AI tools. On the one hand, The New York Times, as well as many other press organizations, has begun to examine how it can use AI and even recruited a manager last year to be responsible for reviewing initiatives on the subject that would help replace some of its workforce with artificial intelligence engines, and create significant savings in personnel costs.
On the other hand, newspapers and media outlets feel very threatened by AI for several reasons:
- Revenue loss: When AI is synchronized in real time with online information, readers will have no reason to access any newspaper or news site. Instead, they will ask the AI questions about the daily news and get an answer, without subscribing to the site, and the newspaper will lose subscribers.
- A decrease in reader traffic to the newspaper's website will hurt revenue from advertising fees, affiliates and other sources of funding that are currently keeping them alive.
- Damage to the brand name: We all encounter so-called "AI hallucinations." That is, completely invented information that AI gives us with great confidence. The Times fears that the AI will associate made-up facts with its real stories, thereby undermining the paper's credibility.
- If newspapers don't find some kind of sustainable reward model, they will quickly go bankrupt, and we will find ourselves with synthesized news that has little to do with truth at best.
Exclusive Benefit
Because of the War, We All Smoking More: The Withdrawal Method That Helped Millions
In association with Alan Carr
There is no doubt that artificial intelligence challenges the legal field a lot, including the question of copyright, but it seems that precisely here, artificial intelligence may receive relatively broad protection in Israel and around the world. Many countries understand that unless AI machines are given the ability to use protected works, they will not be able to fulfill the purpose for which they were created, and we will all lose.
What if this happened in Israel?
As we have seen from several recent cases, it turns out that the law in Israel is advanced relative to other Western countries, at least in terms of its attitude toward innovative technologies.
Already in December 2022, a Justice Ministry opinion clarified that in Israel, in most cases, copyrighted content can be included in machine learning databases. In fact, the Justice Ministry claims that there is no difference between machine learning and human learning – after all, it learns and it learns. The only difference is that in order for the machine to learn, it must copy the work in its entirety, and only at the stage of the product can the changes be examined. This is where the question of fair use comes in.
Fair use? Not sure
Section 19 of the Copyright Act (which corresponds to section 107 of the Copyright Act in U.S. law) is the most significant for the study phase and determines what constitutes "fair use" of protected works. Fair use allows journalists, students, and anyone who studies and enriches their knowledge to use protected works, so a machine should also enjoy this protection.
In both cases, learning is an infrastructure for developing new abilities that encourages development and progress. So it's a bit surprising that The New York Times claims that it itself has copy protections. However, a number of anchors listed in the law may be the ones that tip the scales in favor of The New York Times:
- Purpose of use – The purpose of using protected materials should be for learning and not for commercial purposes.
- Nature of the work – the less protected the work, for example based on facts and data, such as a newspaper article, the fairer its use will be.
- Scope of use of the work – the more the product differs from the original, the fairer it will be deemed.
- The effect of the use on the market of the work – when the aspiration is that the use will not harm the potential deriving from the original work.
The last point raises another dilemma because, on the one hand, there is no doubt that artificial intelligence threatens the newspaper industry. On the other hand, if AI companies were required to pay or obtain approval from every database they want to use, the restriction would actually provide an incentive to engage with massive content entities such as publishers, radio stations, etc., which could harm the level of diversity and objectivity of the databases.
Israel qualifies fair use and explains that there may be cases in which the use will be considered infringing, especially when the machine will make commercial use of the materials from which it learned and sell them at a lower price or compete with the original creators, or when its product will be very similar to the original work.
Although American law is considered the pioneer of "fair use" adopted in Israel as well, no clear guidelines have been published in the United States regarding the question of fair use by a machine. Recent rulings in the United States dealing with copyright, in which a large part of the original work was copied (not by machine), have ruled in favor of the owner of the work. The expected ruling in the New York Times lawsuit will provide an understanding of U.S. policy on the question of fair use by machine.
Finally
Every time significant technology enters the market, it undermines the market and existing models. That's what happened when radio, television, streaming services and now AI came in. As of today, there is no telling what the outcome of the trial will be and it seems that both sides have justified arguments, but what is certain, is going to be interesting.
Iris Zarzaevsky is a strategic consultant to organizations.
Adv. Shani Mizrahi is the owner of a strategy, PR and content firm.
- More on the subject:
- artificial intelligence
- AI
- Press
- New York Times
- CHATGPT