Jaap Ariens/Nurfoto/Getty Images
The burgeoning AI industry reaches another major milestone as two new class-action lawsuits question whether the technology violates privacy rights, collects intellectual property without consent and harms the general public. We have just welcomed Experts believe it is likely to be the first in a series of legal challenges against companies developing such products.
Both lawsuits were filed on Wednesday over the ChatGPT software, a “large-scale language model” that can generate human-like responses to text input, by OpenAI, a research institute comprised of both nonprofits and corporations. is targeted. One, filed by public interest law firm Clarkson, is wide-ranging and poses potential existential threats to AI itself. The other, filed by Joseph Savelli Law Firm and attorney Matthew Buterick, focuses on two prominent authors, Paul Tremblay and Mona Awad, whose books are ChatGPT. The lawsuit claims that it is one of the books that was the subject of training for , and that it is copyright infringement. . (Saveri and Butterick have filed separate lawsuits against OpenAI, GitHub, and Microsoft over their AI-based coding product, GitHub Copilot, claiming that they “violate the terms of the open source license, It seems to be profiting from the work of programmers.”)
OpenAI did not respond to requests for comment regarding the various lawsuits.
Mr. Clarkson’s lengthy filing included a dozen anonymous plaintiffs, beginning with the late theoretical physicist Dr. It will either be for the best or for the worst,” he quotes. He warned of possible “collapse of civilization” if tech companies do not pay attention to the risks of artificial intelligence, and warned that various versions of ChatGPT as well as OpenAI’s public tools, including the image creation product Dall-E He claims to have stolen his personal information. In some cases, he personally identified information from “millions of Internet users,” including children, and “without informed consent or knowledge.” there is The lawsuit seeks a temporary freeze on commercial access and development of OpenAI products until more safety measures are put in place.
The latest lawsuit filed by Saveri and Buterick pursues OpenAI for direct copyright infringement and violation of the Digital Millennium Copyright Act (DMCA). Tremblay (writer of novels) hut at the end of the world) and Awad (author of 13 ways to see fat girls and bunny) represents a group of potential plaintiffs seeking not only injunctive relief in the form of modifications to ChatGPT, but also damages. The filing includes ChatGPT’s detailed responses to user questions about the synopsis of Tremblay and Awad’s book, and is evidence that OpenAI is unfairly profiting from infringing material collected by the chatbot. Lawyers argue that there is.
Lawsuits tread uncharted legal territory, but were more or less inevitable, according to those who study AI technology and privacy, or who practice law related to these issues. .
“[AI companies] Ben Winters, Senior Advisor to the Electronic Privacy Information Center and head of the organization’s AI and Human Rights Project, should have anticipated these kinds of challenges, and may have anticipated them. said to be expensive. He noted that OpenAI CEO Sam Altman referred to past “frivolous” lawsuits against the company during congressional testimony on artificial intelligence in May. “Any time you create a widely used tool for personal purposes that involves so much personal data and is so pernicious, it shocks me that there are no legal issues to be expected.” says Winters. “Especially as third parties allow this kind of unfettered access to integrate with their own systems, consumers are exposed to more personal information, keystrokes and browser clicks in ways they never expected. You’re going to get raw information that isn’t very public, such as activity.”
Ari Reitman, a professor of digital media who teaches about emerging technologies at Carnegie Mellon University Heinz College, said another big reason for the lawsuits against OpenAI is that OpenAI has gone from non-profit status to very high value in 2019. He says that it is because he turned to a commercial company. (In fact, the Clarkson lawsuit alleges that damages claims against plaintiffs were a direct result of this transition.) “As soon as you say, ‘Our company is worth $30 billion,’ your Every publicist will do that,” says Reitman, “people will look at you and say, ‘Okay, now you have a target on your back.'” “It will be the final day for both plaintiff and defense attorneys,” he added, while legal teams “threw the kitchen sink” on the issue to set precedent in a currently obscure area. , of the law expected to be in a contest.
“We think it’s very important to see what rights we have here,” Saberi said. rolling stone. “And obviously the companies aren’t asking for permission. You know, they’re not asking for forgiveness either. I guess I didn’t think so.”
Ryan Clarkson, managing partner of public interest law firm Clarkson, called the lawsuit a warning to AI in general in a statement referring to the class action as a “groundbreaking” federal lawsuit. clarified. “OpenAI and Microsoft admit that they do not fully understand their core technology.
They have ignited an arms race,” he said in a statement. “They released it to the world anyway, and it’s rapidly becoming entwined in all aspects of our lives.” (Microsoft has invested billions in OpenAI, and one of its co-founders, It also received millions of dollars in funding from Elon Musk.)
Metab Khan, Resident Fellow at Yale Law School and director of the Yale/Wikimedia Initiative on Intermediaries and Information, says companies using large-scale language models face the legal challenges ahead. He said it was unclear whether he was prepared enough for the scope of the problem. “I believe that common practice in LLM development has largely ignored copyright implications, and that strong measures have not yet been developed that consider user rights and copyright concerns,” she said. say. Still, she has questions about the Clarkson lawsuit, which she says “seems to have bundled together” many individual and topical issues into one complaint. She hopes to see more lawsuits related to copyright lawsuits and computer fraud and abuse laws that “prohibit intentional computer access without or beyond permission.” That would probably raise questions about not only the defamation claim, but also OpenAI’s immunity under Section 230 (a clause that protects the provider of Internet services from liability for what users say and what they publish on the platform). Deaf, she says.
Regarding the copyright lawsuit filed by Saberi and Buttarik, Khan considered the plaintiffs’ claims that the books were included in ChatGPT’s source material to be “thinly grounded”, noting that “the data set It’s an incomplete picture of what is being developed and trained.” She admits that OpenAI reveals little about these datasets, but that the authors believe their copyrights have been infringed and that “their work and the output generated by the chatbot It is necessary to prove both “substantial similarity with
Noah Downs, an intellectual property attorney and partner at Premack Rogers, said both lawsuits “look pretty bad,” but Clarkson’s chances of winning are higher than those of Saberi and Butterrick. is also high. We take fairly objective measures against infringement. Copyright claims, on the other hand, are “more of an argument” and a “theoretical argument.” Both he and Khan expect OpenAI to opt for the “classic fair use defense” in this matter, which Downes explained is that “usually if a defendant is intended to comment, criticize, or comment when used without permission.” It’s about educating others about the underlying work. “Despite the various merits of the two arguments,” each lawsuit has enough claims that OpenAI cannot overcome them all, and each at least he is ultimately responsible for one offense. It is expected that there will be more,” he said.
On the privacy front, the lack of strong privacy protections in the United States has led big tech companies to avoid liability for intrusive data practices, Winters said. “They are likely to use the First Amendment protections regarding the scraping and use of personal data in conjunction with Section 230 to deny liability,” he said. Reitman said US citizens are more at risk on this front because there is no such thing as a strong General Data Protection Regulation in the European Union, and OpenAI only uses publicly available online information. I agree that it can be argued that
“However, the fact that a company does this and provides access to this information in a way that I have not consented to [plaintiffs’ attorneys are] I’m going to argue that it violates privacy laws and violates privacy,” Reitman said. He compares it to what we’ve seen with social media giants like Facebook. “It repeats exactly the same when it comes to how to access data and harness surveillance capitalism,” he observes. ”
There are no shortage of angles to attack OpenAI, but Butterrick, a Hollywood designer and programmer who returned to the legal practice because of the problems posed by AI, said: rolling stone Protecting creative property is a worthwhile approach. “When people start entering data into websites and talking about privacy, the issue of terms of service comes up,” he says. “Legally, the authors of the books do not have a contract with OpenAI. I know. It’s not legal. Legally speaking, I think this is a nice clean fact.”
Butterrick and Saberi are hesitant to predict an avalanche of lawsuits against AI companies, but they sense a shift in public mood, and Butterrick said the specter of this technology could be seen in entertainment. It mentions how it has galvanized the strike movement in the industry. Saberi laughs, recalling that OpenAI’s CEO previously dismissed GitHub’s complaint as baseless, even though a federal district court found it sufficient to proceed. I was. “It’s kind of rough, isn’t it? Just kidding,” he says. “There is a slight gap.”
It’s only been six months since the first OpenAI lawsuit, but “it feels like a lot has changed,” Saberi said. “We were really in uncharted territory.” Since then, there have been far more public debates, including US congressional hearings, and more momentum to pass AI regulations in Europe, he notes. “And while people are thinking about it and worrying about it, they have found a way to reach our boundaries.”
