OpenAI is once again in the legal spotlight. A collective of writers, including prominent figures like Michael Chabon and David Henry Hwang, have filed a lawsuit against the tech giant. The core of their complaint is they believe OpenAI unlawfully utilized their copyrighted works to train its AI model, ChatGPT. Chabon and the group have also brought a similar lawsuit against Meta Inc. for the same reasons.
ChatGPT Access Work without Permission
The lawsuit, which seeks class-action status, emphasizes that ChatGPT's capability to summarize and analyze content penned by these authors is a clear indication that OpenAI trained its GPT large language model using their works.
The plaintiffs argue that the outputs generated by ChatGPT are essentially “derivative” works, which directly infringe on their copyrights. The complaint goes on to state, “OpenAI's acts of copyright infringement have been intentional, willful, and in callous disregard of Plaintiffs' and Class members' rights.”
It's worth noting that Michael Chabon, known for his book “The Amazing Adventures of Kavalier & Clay,” was among the 10,000+ authors who previously signed an open letter urging tech companies, including OpenAI, Meta, and Google, to seek consent and provide fair compensation to authors when using their works for AI training.
A Pattern of Disputes
This isn't the first time OpenAI has been challenged on such grounds. Earlier in the year, Sarah Silverman, Christopher Golden, and Richard Kadrey accused both OpenAI and Meta of copyright infringement.
They claim that the companies obtained their books from illegal sources, such as websites that offer free downloads of pirated books. They name Bibliotik, Library Genesis, Z-Library, and others as examples of such websites. They say that their books were available on these websites and were downloaded in large quantities by the companies or their partners.
Similarly, authors Paul Tremblay and Mona Awad filed a lawsuit against OpenAI in June. The current lawsuit not only demands compensation for the alleged copyright violations but also urges the court to prevent OpenAI from continuing what they deem as “unlawful and unfair business practices.”
In July, a group of leading news publishers also considered suing AI companies over copyright infringement. The publishers allege that the AI firms are infringing on their intellectual property rights and undermining their business model by scraping, summarizing, or rewriting their articles and distributing them on various platforms, such as websites, apps, or social media.
Privacy Concerns and Data Misuse
Apart from copyright issues, OpenAI is also facing allegations related to privacy breaches. Another class-action lawsuit suggests that OpenAI's machine learning models, including ChatGPT and DALL-E, have been collecting personal data from the internet, violating several privacy laws. This data collection is believed to encompass users' images, locations, music preferences, financial details, and private communications, especially through integrations with platforms like Snapchat, Spotify, Stripe, Slack, and others.
The new lawsuit emphasizes that such data collection breaches the terms of service of these platforms and privacy laws, constituting unauthorized access to individuals' information.
The outcomes of these lawsuits could have far-reaching implications for the AI industry. If the court rules in favor of the plaintiffs, it could reshape the regulatory landscape around AI, copyright, and privacy. This could lead to stricter rules, compliance requirements, and potential financial penalties for OpenAI, affecting its financial stability and fundraising capabilities. Moreover, companies using OpenAI's products might reconsider their associations to safeguard their reputation and ensure users' privacy.
The legal challenges against OpenAI are part of a broader debate on the ethics and legality of AI training data. The question of whether using data from the internet for AI training constitutes “fair use” under copyright law remains unresolved. Some legal experts argue that if the AI-generated content is transformative and significantly different from the original, it might be considered fair use. However, the line between emulating style and direct copying remains blurred, especially when original elements from artworks appear in AI-generated images.