OpenAI responded to The New York Times' lawsuit accusing the newspaper of manipulation
Mariia Shalabaieva/Unsplash
OpenAI has published a response to a copyright infringement lawsuit filed against it by The New York Times. Representatives of the startup called the publication's claims groundless.
Here's What We Know
In a statement on its blog, OpenAI says it is already working with a number of news organisations and creating new opportunities for them. The company considers training AI models on other people's content "fair use" but grants publishers the right to ban such use as a sign of respect.
In addition, OpenAI said the problem of repeating whole sections of specific articles is a rare error that engineers try to minimise. It most often occurs when identical text fragments are included in the training set.
According to the company, until 19 December, negotiations with The New York Times were constructive and concerned the display of the newspaper's content with attribution in ChatGPT. However, the lawsuit of 27 December, which OpenAI learned about from the media, came as a surprise.
The developers got acquainted with the claims and accused the publication of manipulation. According to them, the examples of violations from the Times article could be reproduced with only 10-20 attempts. Also in the examples NYT used its old articles, which were reprinted by many Internet resources, which caused the effect of repetition.
Despite the lawsuit, OpenAI has expressed its willingness to continue negotiating a partnership.
Go Deeper:
- The New York Times is suing OpenAI and Microsoft over its use of the publication's content to train AI
- NYT banned from using its content to train generative artificial intelligence
Source: OpenAI