In December, the New York Times filed a lawsuit against OpenAI—the creator of the popular artificial intelligence software known as ChatGPT—and Microsoft, one of OpenAI’s primary financial backers and partners. As reported by the Times, the suit alleged that OpenAI used millions of Times articles to train “automated chatbots that now compete with the news outlet as a source of reliable information” by reproducing Times articles. By doing this, the suit argued, OpenAI was trying to “free-ride” on the newspaper’s investment in journalism, and Microsoft was guilty of doing the same, because it used ChatGPT technology in its Bing search engine and in many Microsoft Office products. The Times didn’t ask for specific financial damages from OpenAI or Microsoft, but said that the behavior alleged in the lawsuit should result in “billions of dollars” in damages from both companies. The Times also asked the court to force OpenAI to destroy any AI models, databases, and training data that were based on copyrighted material from the paper.
As I reported for CJR in January, OpenAI’s response to the lawsuit was twofold: On the one hand, it argued that the Times was not being transparent about the process it had used to get ChatGPT to produce the copies of Times articles, and that getting ChatGPT to do this involved a bug that users would likely never experience. At the same time, OpenAI argued that its scanning or “ingestion” of data from sources such as the Times to feed its AI engine was permissible under the fair use exemption in American copyright law. As I explained in October, according to the fair use principle, copyrighted material can be used for certain purposes without permission, and without paying the owner a fee for licensing, provided the use meets certain criteria (as outlined in the “four factors” test that judges use when hearing fair use cases).
Last week, OpenAI filed an official response to the Times lawsuit that echoes and expands on many of the arguments the company originally made in January. Despite the allegations that ChatGPT could become a competitor to the Times by reproducing articles, OpenAI said in its response that ChatGPT is “not in any way a substitute for a subscription to the New York Times.” In the real world, OpenAI said, people not only do not use ChatGPT for that purpose, but would not be able to do so even if they wanted to, since “in the ordinary course, one cannot use ChatGPT to serve up Times articles at will.” According to OpenAI, the allegations in the Times’ lawsuit “do not meet its famously rigorous journalistic standards.” In reality, the company said, the Times paid someone to hack OpenAI’s products, and it took this person or persons tens of thousands of attempts to generate the kinds of results included in the suit.
Note: This was originally published as the daily email newsletter for the Columbia Journalism Review, where I am the chief digital writer
OpenAI also noted in its response that it has established “important partnerships” with many leading news companies, including the Associated Press and Axel Springer, the German media giant that owns Insider (formerly known as Business Insider) as well as Politico and German outlets such as Bild. These collaborations, OpenAI says, have allowed the company and its partners to “creatively explore and implement AI solutions that assist investigative reporting, create enhanced reader experiences, and improve business operations.” OpenAI added that the Times’ suggestion that journalism would somehow be imperiled by the development of artificial intelligence was “pure fiction.”
On Monday, Microsoft also filed papers responding to the lawsuit and asking a judge to throw out several of the claims against the company. The statement opened with a description of how, in 1982, Jack Valenti, then the head of the Motion Picture Association of America, told Congress that a new technology known as the videocassette recorder should be stopped because “the VCR is to the American film producer and the American public as the Boston Strangler is to the woman at home alone.” In a similar way, Microsoft argues in its motion to dismiss the lawsuit that the Times is using “its might and its megaphone to challenge the latest profound technological advance,” namely the large language model, or LLM. Copyright law, Microsoft says, should represent no more of an obstacle to the development of artificial intelligence engines than it did to the VCR “or the player piano, copy machine, personal computer, internet, or search engine.”
In its brief, Microsoft asks the court to throw out three specific claims: that Microsoft is liable for copyright infringement by end users of its tools; that the company has broken the law by avoiding copyright protections, which is an offense under the Digital Millennium Copyright Act; and that ChatGPT misappropriates content in a way that is illegal under state laws designed to protect “hot news,” or time-sensitive scoops. Microsoft argues that these claims are not applicable or that the facts supporting them are unproven: the Times has shown no evidence of end-user infringement of its content, Microsoft says, only infringement by the newspaper’s own lawyers. Additionally, Microsoft says that no evidence is provided that either it or ChatGPT evaded copyright protections under the DMCA, and that the “hot news” doctrine that exists in common law is irrelevant because it is superseded by the federal Copyright Act, which provides fair use as an exception to infringement.
The arguments about how the Times crafted its prompt to output verbatim copies of its articles, or about the “hot news” doctrine and how it might apply to the case, are to some extent sideshows in this lawsuit. As I explained in October, the core of OpenAI’s defense against the Times—and, by extension, Microsoft’s defense as well—is that whatever scraping or ingesting of content OpenAI did was covered by the fair use provision of US copyright law. OpenAI says that a long history of court decisions have upheld the idea that it is legal to use copyrighted content as part of a process that “results in the creation of new, different, and innovative products” such as ChatGPT. The company argues that no one should be able to monopolize facts, and that the Times can’t stop AI models from acquiring these facts, any more than other news organizations can prevent the Times from “re-reporting stories that it had no role in investigating.”
The Times‘ lawsuit isn’t the only legal battle OpenAI is facing. A number of authors and other content creators have sued the company because their work was scraped by OpenAI without their permission. But last week, OpenAI was hit by a very different kind of lawsuit from Elon Musk, who in addition to owning X was a cofounder and early financial backer of OpenAI. In his claim, Musk argues that OpenAI and Sam Altman, its founding chief executive officer, violated the company’s original mission by putting the pursuit of profit ahead of the use of AI to benefit humanity. Musk says that he and Altman agreed when they created it that OpenAI would be—as its name suggests—an open-source nonprofit that would develop AI “for the benefit of humanity, not for a for-profit company seeking to maximize shareholder profits.”
Until recently, Musk argues, OpenAI sold the rights to its AI technology to Microsoft but at the same time released the code behind that tech as open source, so that others could develop their own AI engines. Musk says this changed with the latest version of OpenAI’s software, known as GPT-4, which was given to Microsoft alone and now (Musk alleges) forms the basis of the artificial intelligence inside products such as Microsoft Office. This change, Musk says, came after a turbulent period at OpenAI (which I wrote about for CJR) in which two board members tried to oust Altman but instead were removed themselves. Musk argues that this consolidated Altman’s power over the company. According to Bloomberg, the Securities and Exchange Commission is investigating whether the company misled its investors during that period; other regulatory agencies, including the European Union and multiple US authorities, are said to be looking into the relationship between OpenAI and Microsoft.
One interesting aspect of Musk’s lawsuit is that he argues that Microsoft should not have been allowed to license the latest version of OpenAI’s software because GPT-4 is too close to achieving what is known as “artificial general intelligence,” or the ability to reason in much the same way a human being would. Musk says Microsoft’s deal with OpenAI only gave the company access to products that had not shown AGI-level capabilities, which GPT-4 allegedly has. According to Musk, researchers at Microsoft have publicly stated that they believe GPT-4 could “reasonably be viewed as an early (yet still incomplete) version of an artificial general intelligence (AGI) system,” although there is some debate within Microsoft about whether these claims are true. Musk also said that it is his belief that OpenAI is developing a new AI model known as Q* (pronounced “Q star”) that is even more likely to demonstrate artificial general intelligence.
On Tuesday, Altman and other senior executives at OpenAI responded to Musk’s lawsuit, saying the company would be moving to dismiss all of his claims. They also noted that in the early days of OpenAI, after it became clear that the company would need a lot of money to pay for the computing power that AI requires, Musk wanted majority equity and board control—and also wanted to be CEO. According to Altman and the other executives, at one point Musk suggested merging OpenAI into Tesla, his electric-car company. The statement says that the board felt it would be “against the mission for any individual to have absolute control over OpenAI,” adding that the company is “sad that it’s come to this with someone whom we’ve deeply admired—someone who inspired us to aim higher, then told us we would fail, started a competitor, and then sued us when we started making meaningful progress toward OpenAI’s mission without him.”