OpenAI Responds to New York Times Lawsuit, Claims Paper “Intentionally Manipulated” Prompts
Jan 08, 2024 - hollywoodreporter.com
OpenAI is defending itself against a copyright infringement lawsuit from The New York Times, which alleges the AI company used the publisher's articles to train its chatbot. In a blog post, OpenAI claimed the Times "intentionally manipulated" prompts to make it appear as if their AI, ChatGPT, was generating near word-for-word excerpts of articles, which they insist is a "rare bug". The company also stated that there are measures in place to prevent "inadvertent memorization" and users are prohibited from prompting models to produce answers that may infringe on intellectual property rights.
The lawsuit was filed in response to evidence presented by the Times of OpenAI and Microsoft's products displaying near word-for-word excerpts of articles when prompted. This allegedly goes beyond the snippets of texts typically shown with ordinary search results. OpenAI, however, argues that using the publisher’s articles and other “publicly available internet materials” for training AI models is fair use.
Key takeaways:
OpenAI is defending itself against a lawsuit from The New York Times, which accused the company of copyright infringement for using its articles to train its AI chatbot.
The company claims that the Times is not providing the full context and that it manipulated prompts to make it seem like the chatbot was copying articles verbatim.
OpenAI insists that such verbatim copying is a rare bug and that there are measures in place to prevent inadvertent memorization.
The company argues that using publicly available internet materials, including the Times' articles, to train AI models is fair use.
