OpenAI has firmly responded to The New York Times' copyright lawsuit, dismissing the allegations as "without merit" and expressing optimism about the possibility of a future partnership with the renowned media outlet.
Addressing the accusations in a blog post, OpenAI contended that the Times failed to present the complete narrative, specifically challenging the claim that its ChatGPT AI tool replicated Times articles verbatim. OpenAI argued that the Times manipulated prompts to include excerpts from articles, asserting that their models do not typically behave in the manner suggested by the media outlet. The company suggested that the Times either instructed the model to regurgitate content or selectively chose examples from multiple attempts.
OpenAI maintained that it has actively worked to minimize content duplication in its large language models. Additionally, the company claimed that the Times declined to provide examples of the alleged reproduction before filing the lawsuit. OpenAI acknowledged that it removed a ChatGPT feature, called Browse, which inadvertently reproduced content.
Despite these developments, OpenAI reiterated its longstanding position that AI models require access to a vast pool of human knowledge for effective learning and problem-solving. While acknowledging the legal rights of copyrighted works and offering opt-outs to training data inclusion, the company argued that training AI models with internet data falls under fair use rules that permit the repurposing of copyrighted works.
OpenAI emphasized its commitment to respecting the diverse spectrum of human intelligence and experience, maintaining that AI systems like ChatGPT cannot be developed without access to copyrighted content. The company announced the option for website owners to block its web crawlers from accessing their data starting in August 2023, almost a year after ChatGPT's launch.