The startup specializing in artificial intelligence knowledge asserted that it collaborated with media agencies and refuted The Times’ accusation of copyright infringement, suggesting that the newspaper’s allegations were incomplete.
Engaging in a discourse on the unauthorized utilization of published material for instructing AI technologies, OpenAI stated on Monday that The New York Times’ legal action against it held no significance and emphasized its support for media enterprises.
On December 27, The Times initiated a legal challenge against OpenAI and Microsoft, alleging that the latter had breached its copyrights by utilizing a substantial amount of its content to train AI models like the ChatGPT bot. The lawsuit contended that these AI systems were now competing with The Times as a reliable source of information.
In a 1,000-word statement released on Monday, OpenAI asserted that it collaborated with media organizations and had established partnerships with some, including The Associated Press. The company argued that employing copyrighted material for enhancing its technology constituted fair use. However, The Times claimed that the lawsuit failed to provide a comprehensive understanding of OpenAI’s operations and technologies.
“We envision further collaborations with media entities to enhance their content creation capabilities by harnessing the transformative power of AI,” OpenAI stated.
When approached for additional comments, OpenAI representative Lindsey Held chose not to elaborate.
The lawsuit filed by The Times against OpenAI and Microsoft for copyright infringement marked a significant legal action by a prominent British media outlet. Various AI firms have faced similar lawsuits from diverse sectors, such as literature and software development, due to the surge in “generative AI” technologies that generate text, images, and multimedia content based on brief prompts.
OpenAI, along with other AI companies, has developed this technology by feeding vast amounts of online data, some of which may be copyrighted. This practice has underscored the untapped potential of online content, ranging from stories and news articles to forum posts and images.
Given that AI systems do not replicate content entirely and operate on open data, these companies have argued that utilizing such material for training purposes does not necessitate licensing fees.
In a published statement, OpenAI mentioned positive discussions with The Times regarding potential collaborations, with the most recent exchange occurring on December 19. The company revealed that The Times had observed instances where OpenAI’s technology seemingly reproduced excerpts from its articles verbatim, though specific examples were not provided. OpenAI expressed surprise and disappointment when The Times proceeded with the legal action eight days later.
The Times did not immediately respond to requests for comments.
OpenAI acknowledged that its technology occasionally replicated content, attributing it to a “rare bug” under investigation for resolution. Instances highlighted in ChatGPT’s dispute with The Times indicated direct replication of content from the newspaper’s articles.
According to OpenAI, deliberately manipulating their models to reproduce content was deemed inappropriate usage and violated their terms of service.
