A California-based law firm filed a class-action lawsuit against OpenAI, accusing the AI company of using data scraped from the internet while training its well-known chatbot, ChatGPT, for serious infringement of countless people's copyrights and privacy.
The lawsuit seeks to test a new legal theory that OpenAI infringes millions of internet users when it uses user data such as social media comments, blog posts, Wikipedia entries, and family recipes. The lawsuit law firm Clarkson has previously filed a massive class action lawsuit over issues such as data breaches and false advertising. Ryan Clarkson, the firm's managing partner, said they wanted to represent "real individuals whose information was stolen and commercially exploited to create this very powerful technology."
The case was filed Wednesday morning in federal court in the Northern District of California. A spokesperson for OpenAI did not respond to a request for comment.
The lawsuit involves an important unresolved issue related to the proliferation of "generative" AI tools such as chatbots and image generators. The technology works by ingesting billions of words from the open internet and learning the reasoning relationships between them. After digesting enough data, the resulting "large language model" can predict how to respond to prompts, allowing it to write poetry, conduct complex conversations, and pass professional exams. But the people who wrote those billions of words never agreed to let companies like OpenAI use them for profit.
"All of this information was acquired on a large scale and was never intended to be used to train large language models," Clarkson said. He expressed hope that the courts will develop some norms on how AI algorithms are trained and how people who use their data are compensated.
The firm already has a group of plaintiffs and is actively seeking more.
The legality of using data obtained from the public internet to train tools with high profit potential for developers remains unclear. Some AI developers argue that data obtained from the Internet should be considered "fair use," a concept in copyright law that creates exceptions if material is altered in a "transformative" way. Intellectual property lawyer Katherine Gardner said: "The issue of fair use is an unresolved issue that we will see resolved in court in the coming months and years. Artists and other creative professionals who can show their copyrighted work being used to train AI models may dispute the companies that use the work, but she said it may be difficult for those who only post or comment on the site to get compensation.
"When you post content on social media or any website, you usually grant the site a very broad license right to use your content in any way," Gardner said. "It will be very difficult for the average user to claim that they are entitled to any form of payment or compensation for the use of their data," she said. ”
The lawsuit also adds to a growing list of legal challenges faced by companies looking to profit from AI technology. In November, a class-action lawsuit was filed against OpenAI and Microsoft's use of computer code training tools from GitHub, a Microsoft-owned online coding platform. In February, Getty Images filed a lawsuit against Stability AI, a smaller AI startup, alleging that it illegally used its photos to train its image-generating robots. And this month, OpenAI was sued by a Georgia broadcaster for defamation, saying ChatGPT-generated text errors accused him of fraud.
OpenAI isn't the only company using massive amounts of data scraped from the open internet to train its AI models. Google, Facebook, Microsoft, and a growing number of other companies are doing the same. But Clarkson chose to target OpenAI because it captured the public imagination on ChatGPT last year, inspiring its larger rival to launch its own AI technology.
"They are the companies that set off this AI race," he said. They are nature's primary goals. ”
OpenAI has not disclosed what data is used in its latest model, GPT4, but previous versions of the technology have been confirmed to incorporate Wikipedia pages, news articles, and social media comments. Chatbots from Google and other companies have used similar datasets.
Regulators are discussing enacting new laws that would require companies to be more transparent about the data they use in AI. Gardner, an intellectual property lawyer, said a court case could also prompt judges to compel companies such as OpenAI to provide information about the data used.
Some companies are trying to stop AI companies from scraping their data. According to the Financial Times, music distributor Universal Music Group had asked Apple and Spotify to block the crawler. Social media site Reddit is shutting down its data stream access, citing that big tech companies have been crawling comments and conversations on its site for years. Twitter owner Elon Musk has threatened to sue Microsoft, saying it used data obtained from Twitter to train its artificial intelligence. Musk is building his own artificial intelligence company.
The new class-action lawsuit against OpenAI further alleges that the company is not transparent enough with users who use its tools, and that data fed into the model could be used to train new products that the company will profit from, such as the Plugins tool. The lawsuit also alleges that OpenAI doesn't do enough to ensure that children under the age of 13 don't use its tools, something other tech companies, including Facebook and YouTube, have been accused of for years.
Follow me and don't get lost!