Amid lawsuits from media companies accusing OpenAI’s ChatGPT, as well as other Large Language Models, of using copyrighted news content without giving credit or compensation, the AI firm has been attempting to mitigate compromises with media firms in order to avoid shutting down.
Sam Altman, the CEO of OpenAI, who recently returned to the company after a temporary coup by its board, has reportedly been participating in talks with outlets such as CNN, Fox Corp, etc.. The goal is to negotiate deals to pay for access to their content in order to continue training their developing chatbot. The company has already announced deals with the Associated Press and Axel Springer over content uses.
Yet despite these negotiations, OpenAI remains in active legal disputes, the most pressing being recently filed by the New York Times– and experts have speculated that these collective litigations form a strong case for OpenAI to either stop stealing news content or give considerable compensation for it.
The startup’s separatist approach to these negotiations generates the risk of exposing it to conflict as U.S. lawmakers and federal courts examine the legality of AI training; compromising with individual entities makes them liable for how each one acts differently, a less predictable outcome than one resulting from an expansive negotiation with all outlets combined.
These separate deals are what prompted the Times’ lawsuit in the first place; the news outlet reported that they opted to sue OpenAI after talks on potentially reaching an “amicable solution,” such as a licensing deal, were dissolved.
The Times’ lawsuit is still likely years away from a court date, and digital papers run the high risk of having their content stolen by chatbots until significant legal parameters are set in place. The race to achieve super-efficient AI is in full swing, after all; the likelihood that companies will wait in limbo for legislation to pass was already small, and the fierce level of competition doesn’t help.
While OpenAI described the suit as “without merit” in a Jan. 8 blog post and alleged that the Times “intentionally manipulated prompts, often including lengthy excerpts of articles, in order to get our model to regurgitate”. The Times maintains they have a significant case, holding proof of near-perfect copying done by ChatGPT, according to James Grimmelman, professor of digital and information law at Cornell Law School.
Kristelia Garcia, a copyright law expert and professor at Georgetown University Law Center, claimed that it would be surprising if OpenAI even allowed them to go to trial, given the extensive amount of financial and legal stakes in play. She also alleged that the enormous statutory damages that could arise out of this lawsuit would not only stop OpenAI’s business models but probably shut their startups down completely.
While The Times is just one of the many media companies fielding lawsuits against OpenAI, it has been the most vocal and open about its activity; others prefer to remain unnamed, as reported by Bloomberg.
