A court ruling by the federal court could put OpenAI in serious legal and financial danger in a landmark copyright infringement case.
As reported by Bloomberg, the AI giant has been ordered by the U.S. District Court of the Southern District of New York to provide internal communications, such as Slack messages and emails, regarding its deletion of a dataset with pirated books. According to legal experts, this could translate to billions of dollars in damages against the company.
The plaintiffs, a committee of authors and publishers, argued that the internal conversations of OpenAI cannot be regarded as the confidential information of an attorney-client.
They are seeking evidence of the existence of “willful infringement,” which would raise the statutory damages to up to $150,000 per copyrighted work.
The potential liability is staggering due to the tens of millions of books and articles involved in the consolidated lawsuits.
As described by David Schultz, a professor at Hamline University, the stakes go way beyond a large award in damages.
“Finding out what attorneys said or what clients said to attorneys gives us a lot of evidence regarding state of mind.” Such a disclosure would be an "enormous" blow to OpenAI’s defense.
The case reflects a recent lawsuit against Anthropic PBC, which settled in August for a record $1.5 billion under the inordinate pressure of a potential $1 trillion damages judgment.
According to legal analysts, the pressure on OpenAI is building, and it may resort to the same settlement.
Plaintiffs are using a new legal tactic, arguing that OpenAI is waiving privilege by only partially disclosing why it deleted the information.
They are also citing the exception of crime fraud, which is not very common. They claim that the communications can expose criminal infraction or willful tampering with evidence.
If the court finds that OpenAI destroyed evidence in anticipation of a lawsuit, it might impose sanctions such as monetary fines, limited defenses, or a default judgment against the plaintiffs.
This decision would set a landmark precedent for the dozens of copyright cases awaiting trial against AI companies due to their large language model training.