Alec Radford, a researcher who helped develop several core AI technologies at OpenAI, has been subpoenaed in a copyright case against the company. Court documents filed Tuesday in the U.S. District Court for the Northern District of California show Radford received the subpoena on February 25th.
Radford left OpenAI late last year to pursue independent research. He was the lead author of OpenAI's groundbreaking research paper on "Generative Pre-trained Transformer (GPT)," the technology underlying one of OpenAI's most popular products, the AI chatbot ChatGPT. Radford joined OpenAI in 2016, just a year after the company's founding. During his tenure, he contributed to the development of multiple GPT series models, as well as the speech recognition model Whisper and the image generation model DALL-E.
The copyright case, titled "Related to OpenAI ChatGPT Litigation," was filed by a group of book authors, including Paul Tremblay, Sarah Silverman, and Michael Chabon. They allege that OpenAI infringed their copyrights when training its AI models and accuse ChatGPT of failing to properly attribute their works.
Last year, the court dismissed two of the plaintiffs' claims against OpenAI, but allowed the claim for direct infringement to proceed. OpenAI maintains that its use of copyrighted data for training falls under fair use.
Besides Radford, the authors' legal team has attempted to subpoena several other prominent figures, particularly two former OpenAI employees, Dario Amodei and Ben Mann, who are known for leaving the company to found Anthropic. Amodei and Mann objected to the subpoenas, arguing that the requests were overly burdensome.
This week, a U.S. District Judge ruled that Amodei must submit to hours of questioning about his work at OpenAI, relating to two copyright cases, including one brought by the Author's Guild.
Key Points:
📝 Alec Radford, due to his involvement in OpenAI's research, has been subpoenaed in a copyright case.
📚 The copyright case was filed by several book authors, accusing OpenAI of unauthorized use of their works for AI training.
⚖️ OpenAI maintains that its use of training data constitutes fair use, while the court allows the claim for direct infringement to proceed.