Over the past year, numerous copyright holders have sued AI developers, accusing them of using copyrighted material without permission to train their AI models.
Some of the lawsuits come from the music industry. In such a case, universal music group, Concord Music Groupand ABCCO Sue artificial intelligence developers Anthropic selection Allegedly using copyrighted lyrics to train its chatbot Cloud.
Perhaps most noteworthy are the record companies owned by the Big Three— Sony Music GroupUniversal Music Group, and Warner Music Group – filed suit earlier this year sun and sharetwo generative AI music-making platforms that can create a track with nothing more than a few text prompts, some believe it has become very good at creating music.
The three majors accuse Suno and Udio of infringing the copyright of their recordings by training their artificial intelligence on those recordings, and that the two music generators are now outputting music similar to what they were trained on.
How the courts interpret these cases will be critical for rights holders, including the music industry, as it relates to how the relationship with artificial intelligence will develop in the coming years – especially given the slow progress of legislation regulating artificial intelligence (Although in the right direction, from a rights holder’s perspective).
Open artificial intelligencethe super popular creator Chat GPT The chatbot has been subject to multiple lawsuits from news organizations accusing OpenAI of using its copyrighted news articles to train ChatGPT. One of the most famous cases was that of new york timesthe case is still pending in court.
However, another case against OpenAI has been concluded in what could be one of the first rulings on the relationship between artificial intelligence training and copyrighted material. On the face of it, this doesn’t look good for the copyright holder: OpenAI won the case.
Earlier this year, original storya progressive-leaning online news site, and affiliates altenetsued OpenAI, asking the New York Federal Court to compensate for losses caused by OpenAI’s alleged stripping of content management information. original story and altenet articles and use them to train artificial intelligence. The news organizations are also asking for an injunction to prevent OpenAI from using its content in future training.
In a ruling on Thursday (November 7), the judge Colleen McMahon The U.S. District Court for the Southern District of New York ruled in favor of OpenAI and dismissed the lawsuit filed by OpenAI original story and altenet.
The judge’s reasoning? These news organizations have been unable to prove that OpenAI caused them any actual harm by using their content.
“I do not believe there are any historical or common law parallels in merely removing identifying information from a copyrighted work — without disseminating it —” Judge McMahon wrote in her dismissal. The full article can be read here.
Judge McMahon concluded that given that ChatGPT was trained on countless pieces of data, it was actually unlikely that it could repeat the copyrighted article or substantial portions of it in response to the questions posed by users.
This isn’t necessarily the end for news organizations: The judge dismissed the case “without prejudice,” meaning original story If they can make a stronger argument that they suffered harm, they could resubmit their case to the court, although the judge did write that she was “sceptical about the plaintiff’s ability to claim identifiable harm.”
Nonetheless, the lawyer matt theme of Lowe + Lowewhich represents original story In this case, tell Reuters He is “confident that we can resolve the issues raised by the court through an amended complaint.”
In other words, this case is not completely over yet.
“I do not believe there are any historical or common law parallels in merely removing identifying information from a copyrighted work — without disseminating it.”
Judge Colleen McMahon, Raw Story v. OpenMedia
At the same time, however, some observers believe this is bad news for copyright holders. Drew Thurlowformer A&R director Sony Music Entertainment Now the founder of a music startup Opening Ceremony Mediaargued in a LinkedIn post that this could strengthen Suno and Udio’s argument that their use of copyrighted songs to train AI constitutes “fair use” under U.S. copyright law.
“What is one of the tenets of copyright infringement? The perpetrator must harm the plaintiff’s market and/or financial position,” Thurlow wrote.
“Are these Gen AI companies hurting the recorded music market? So far, certainly not. In fact, there is evidence that consumer AI tools are increasing music engagement. They may even help grow the recorded music market.
If Thurlow’s assertion is correct, then one of the pillars of the fair use doctrine may favor AI companies: that is, it may be difficult for record labels to prove that they or the music market were harmed by AI companies’ use of their Material.
However, Raw Story and OpenAI The case is significantly different from lawsuits filed by the music industry against Anthropic, Suno and Udio. Below, we break down these differences and how they could lead to different outcomes for the music industry’s lawsuits against AI developers.
No copyright issue in Raw Story/OpenAI case
Music industry lawsuits against artificial intelligence companies all have one thing in common: They allege copyright infringement (or, more accurately, lots and lots of infringement).
but original story The lawsuit does not allege copyright infringement; it simply alleges that OpenAI violated U.S. Digital Millennium Copyright Act (DMCA), it is prohibited to strip content management information of copyrighted material (for new articles, this will be the name of the news source, author, publication date, copyright information, etc.).
This is an unusual approach, considering other lawsuits brought against AI companies by rights holders, and we can only speculate as to why original story and altenet Infringement will not be claimed at the same time.
In fact, Judge McMahon himself seemed to sniff out a weakness in the case.
“Let’s be clear about what’s really at stake here. The alleged injuries for which the plaintiffs truly seek compensation are not precluded [content management information] training set from Defendants, but instead Defendants used Plaintiffs’ articles to develop ChatGPT without providing compensation to Plaintiffs,” the judge wrote.
The judge added that using the article without permission “is not the type of harm that … is ‘elevated’ by the DMCA.” “Whether there are other statutes or legal theories that would indeed exacerbate this type of harm remains to be seen. But that question is not before the court today. .
Fortunately for rights holders (at least for now), the issue is before the courts in other lawsuits filed by music companies.
A key defense for AI companies—the “fair use” doctrine—has yet to be tested in AI cases
The fact is original storyThe case against OpenAI focuses solely on content moderation messages, meaning the court has yet to rule on a key defense used by AI companies in their battles with rights holders.
This defense is the “fair use” doctrine, which makes it acceptable to use copyrighted material without permission in certain limited circumstances. A simple example is using snippets of news articles from educational textbooks.
Fair use is a key argument in Suno and Udio’s defense against copyright lawsuits brought against them by the recording giants. In fact, they seemed so confident in this defense that they all but admitted to using copyrighted material in their response to the lawsuit.
U.S. courts use a four-factor test to determine whether something is fair use:
- Purpose and Character of Use – Is the copyrighted work used for educational purposes or commercial purposes?
- The nature of the copyrighted work – whether the work is particularly creative and original.
- Quantity and Substantial Content of the Parts Used – How much of the copyrighted work is being used without permission?
- Effect of use on the potential market or value of the copyrighted work.
In his LinkedIn post, Thurlow’s argument relates to the fourth point. If music rights holders cannot prove that their intellectual property rights have been harmed by the use of artificial intelligence in copyrighted works, or that the market has been harmed, then the rights holders’ claims will be weakened.
Music companies may reject this argument. In each case, they argue that music generated by artificial intelligence is a direct competitor to their intellectual property rights in the music market. Whether they can prove it is another matter.
But this factor alone is unlikely to decide these cases. Courts do not take a systematic approach to fair use; these issues are determined on a case-by-case basis, taking into account all four factors.
In their complaints against Suno and Udio, the record companies attacked the “fair use” argument head-on, addressing each of the four factors.
The first factor – the purpose and character of the use – relates to the degree to which the use of copyrighted material is “transformative.” If you add a fragment of a new article to a textbook, it’s quite “transformative”—its form, context, and purpose are very different from the original article.
For Suno and Udio, “the use here is far from transformative, as there is no functional purpose [the AI models] Complaints from record labels about AI platforms say “their purpose is to acquire copyrighted recordings, not to output new, competing music archives.” The complaints can be read in full here and here.
The second factor focuses on the type of copyrighted work allegedly infringed, placing more weight on some works than others. Copyright protection for functional news articles (such as sports scores) is less stringent than for truly completely creative content (such as a new song).
In their complaints against Suno and Udio, the record companies argued that music recordings were the very works protected by copyright.
The third factor relates to the amount of use of the copyrighted work. A small portion of a copyrighted work can be considered “fair use,” but it can be difficult to prove this when the entire copyrighted work has been used.
The label argued that it was “very clear” that Suno and Udio absorbed “the most significant portion” of the copyrighted song, “as [their] For example, be able to re-create some of the most famous musical phrases, choruses and choruses in popular music history.
Actual harm?
This leaves a fourth factor, the only one that maybe–just maybe–we never Raw Story v. OpenAI case. While the judge in that case did not weigh the “fair use” factor, she did indicate that a mere statement of harm to the copyright owner was not enough to bring a lawsuit.
The record company’s complaint against Suno and Udio claims that the platforms’ artificial intelligence-generated music “poses a significant threat to the market and value of copyrighted recordings.” But will the courts simply agree?
In these cases, this may be a real difficulty faced by copyright owners. This question poses a “what if” scenario – what would be the revenue from recorded music if artificial intelligence platforms like Suno and Udio hadn’t emerged? How much does Michael Buble cost? swing If Udio isn’t (allegedly) using it to train its AI music generator, is it worth it? Tricky.
Record companies may have to conduct in-depth market research to show material damage – if such market research currently exists.
At the same time, dismissal Raw Story v. OpenAI There’s no need to keep music owners up at night: This case is different enough, and limited enough in scope, to leave the door open for vastly different decisions in future cases.global music business
