OpenAI Should Turn Over Execs’ Social DMs and Texts, Authors Say

OpenAI Inc. should turn over executives’ social media and text messages—including those of CEO Sam Altman and President Greg Brockman—in discovery, authors accusing the company and partner Microsoft Corp. of copyright infringement told a federal judge.

The OpenAI executives’ posts on X, formerly known as Twitter, make it clear that they all used social media for work and their accounts should be searched for relevant messages, the Authors Guild and author plaintiffs wrote in a Wednesday letter to Magistrate Judge Ona T. Wang of the US District Court for the Southern District of New York.

The dispute marks the latest in a series of discovery fights between OpenAI and plaintiffs in various cases accusing the artificial intelligence behemoth’s chatbot of illegally training on their work, including wrangling over documents such as New York Times reporters’ notes and chatbot inputs.

The Authors Guild, the country’s oldest and largest professional organization representing writers, sued OpenAI in September 2023 and later joined its suit with another brought by nonfiction authors.

OpenAI has dodged handing over executives’ X direct messages by claiming it lacks possession, custody, or control over its employees’ social media accounts and citing a California labor code prohibiting the company from asking for employees’ social media accounts or content, according to the letter. That position is at odds with relevant federal case law, the plaintiffs argued, which has “repeatedly found that employers have control over their employees and can be required to produce documents in their employees’ possession.”

The letter to Wang linked to a handful of X posts it contends demonstrate that Altman, Brockman, Head of Product Nick Turley, and technical staff member Chelsea Voss have used their accounts for work purposes like recruiting and soliciting product feedback, and therefore don’t qualify as “personal” accounts.

OpenAI should also have to collect and search the text messages of Altman, Brockman, and former Chief Technology Officer Mira Murati because each of their accounts is under the company’s control and likely to contain relevant messages, plaintiffs said.

OpenAI’s consistent refusal to even disclose whether its searching employee texts and DMs jeopardizes the timely completion of discovery, the letter said, because the parties need to have informed conversations about searches to streamline the process.

On the same day as the plaintiffs’ letter, OpenAI also wrote to Wang, alleging the plaintiffs are refusing to respond to discovery requests about the harm they’ve supposedly suffered from OpenAI using their works to train ChatGPT.

“The discovery OpenAI seeks goes directly to the harm Plaintiffs claim to have suffered, which is relevant not only to Plaintiffs’ claims for damages but also to OpenAI’s fair use defense,” OpenAI’s letter said.

The parties are scheduled to have a status conference with the court on Oct. 30, and both sides have flagged these discovery issues in their proposed agendas.

Susman Godfrey LLP, Lieff Cabraser Heimann & Bernstein, and Cowan, DeBaets, Abrahams & Sheppard LLP represent the Authors Guild. Latham & Watkins LLP, Keker Van Nest & Peters LLP, and Morrison Foerster LLP represent OpenAI.

The case is Authors Guild et al v. OpenAI Inc., S.D.N.Y., No. 23-cv-08292, 10/22/24.