Key ex-OpenAI researcher subpoenaed in AI copyright case

Alec Radford, a researcher known for his contributions to OpenAI’s key AI technologies, has been subpoenaed in a copyright case against the AI startup, as revealed in a court filing on Tuesday.

The filing, presented by the plaintiffs’ attorney to the U.S. District Court in the Northern District of California, disclosed that Radford was served a subpoena on February 25.

Radford, who departed from OpenAI at the end of last year to pursue independent research, played a significant role in the development of OpenAI’s generative pre-trained transformers (GPTs), including the widely-used ChatGPT platform.

Having joined OpenAI in 2016, Radford contributed to various models within the GPT series, as well as projects like Whisper for speech recognition and DALL-E for image generation.

The copyright case, referred to as “re OpenAI ChatGPT Litigation,” involves authors such as Paul Tremblay, Sarah Silverman, and Michael Chabon alleging that OpenAI violated their copyrights by utilizing their work to train AI models like ChatGPT without proper attribution.

While some of the plaintiffs’ claims against OpenAI were dismissed by the Court last year, the claim for direct infringement is proceeding. OpenAI argues that its use of copyrighted data falls under fair use.

Apart from Radford, the plaintiffs’ attorneys are also seeking to depose Dario Amodei and Benjamin Mann, former OpenAI employees now involved with Anthropic. Amodei and Mann have contested these depositions, citing excessive burden.

A U.S. magistrate judge recently ruled that Amodei must undergo extensive questioning regarding his work at OpenAI in relation to copyright cases, including one filed by the Authors Guild.