OpenAI recently addressed The New York Times’ copyright infringement lawsuit, claiming that it was a difficult and rare occurrence, and that they are actively working to prevent such incidents from happening again. They also affirmed their desire to be allies with various news organizations.
OpenAI clarifies that using copyrighted content, such as news articles, to train their artificial intelligence models falls under fair use, which is exempted in certain cases by US intellectual property law. Although there is ongoing debate surrounding this issue, OpenAI has received support from various groups who interpret the use of data for AI training as a fair practice. Additionally, OpenAI states that they allow news agencies to opt out of their data being used for training, even if it’s not legally required.
Regarding the issue of ChatGPT regurgitating entire articles, OpenAI acknowledges this as a bug and is implementing measures to prevent such outputs. As for the specific example provided by The New York Times, OpenAI claims that the newspaper only mentioned encountering problematic responses from ChatGPT without providing specific examples for verification. OpenAI believes that the mentioned articles were likely old and copied from various websites outside of The New York Times.
TLDR: OpenAI responds to The New York Times’ copyright infringement lawsuit, stating that their use of copyrighted content for AI training falls under fair use. They are actively working to address bugs like ChatGPT regurgitating entire articles. The provided example by The New York Times did not offer verifiable evidence, and OpenAI believes the referenced articles were likely copied from other sources.
Leave a Comment