In a blog post dated January 8, OpenAI responded to a lawsuit initiated by The New York Times (NYT), describing the legal action as unfounded and outlining its collaborative history with various news organizations. OpenAI expressed surprise and disappointment upon learning about the lawsuit from the NYT itself, filed on December 27, despite ongoing discussions between the two parties that seemed to be progressing positively.
The lawsuit, brought against OpenAI and Microsoft by the NYT, accuses them of unauthorized use of the newspaper's content to train AI chatbots. OpenAI's rebuttal challenges this claim, emphasizing the company's commitment to ethical practices and its desire to clarify its operations and technology development approach. OpenAI defended its actions by highlighting four key points: its proactive engagement with news outlets to foster journalism innovation, its belief in 'fair use' of training materials coupled with an 'opt-out' option for publishers, addressing rare instances of content 'regurgitation' in AI outputs, and suggesting that the NYT's portrayal of the situation was incomplete.
OpenAI showcased its media industry partnerships, like the one with German media conglomerate Axel Springer, aimed at mitigating AI-generated inaccuracies or "hallucinations." Additionally, the company mentioned its collaboration with the News/Media Alliance, serving as a platform to address concerns and identify mutually beneficial solutions. This comes after the News/Media Alliance’s critical stance in October, evidenced by a 77-page report and a submission to the U.S. Copyright Office, highlighting the disproportionate use of news content in AI training datasets.
The company also pointed to its 'opt-out' mechanism, a feature that restricts AI tools from accessing content from participating publishers, noting that the NYT itself agreed to implement this option in August 2023. A focal point of the NYT's lawsuit is the alleged over-representation of its website content in OpenAI's training data, surpassed only by Wikipedia and the U.S. Patent Document Database.
The NYT claims that despite reaching out to OpenAI and Microsoft in April 2023 to address intellectual property concerns and seek a peaceful resolution, no agreement was reached. Lawyers have deemed the NYT's lawsuit as one of the strongest cases yet against generative AI for copyright infringement. OpenAI maintains that the alleged misuse of NYT content is not typical or sanctioned, and that its AI does not replace the original NYT content. The company asserts ongoing improvements in its models to prevent training data replication and remains hopeful for a constructive partnership with the NYT, respecting its journalistic legacy. The blog post concludes by reaffirming OpenAI's belief in the lack of merit in the NYT’s lawsuit.
