New York Times and OpenAI Battle Over ChatGPT Training Data
OpenAI in Bitter Dispute with the New York Times
The New York Times (NYT) has challenged OpenAI's dismissal of its lawsuit, which accuses OpenAI and Microsoft of using its articles without permission. OpenAI claimed the NYT had unfairly tried to trick its AI, ChatGPT, into revealing it had used NYT articles in its training data, calling this a "hack." The NYT disputes this, saying their methods were aimed at uncovering evidence of copyright infringement.
NYT Claims Its Content Was Used Without Permission to Train ChatGPT
OpenAI suggests such uses of ChatGPT by NYT are abnormal, but the NYT counters, pointing out that users often try to use ChatGPT to get around paying for content. The newspaper is trying to determine how many of its articles were used by OpenAI's AI models, a detail OpenAI has not made public. The NYT's actions were in an effort to reveal the extent to which its content has been used without permission, not to set up a lawsuit.
NYT Claims ChatGPT Feature Breached Its Paywall
Furthermore, the NYT highlights that after introducing a new feature that allowed ChatGPT to access current articles, including those behind paywalls, OpenAI briefly disabled this feature due to concerns about copyright infringement. This move was unpopular among some ChatGPT users who enjoyed the feature.
The NYT also points out that before this feature was introduced, they had informed OpenAI of copyright infringement, which OpenAI did not acknowledge in its defense. The NYT argues this situation is similar to past copyright infringement cases, suggesting OpenAI knew of and contributed to the infringement.
OpenAI Is Facing Several Lawsuits Over How ChatGPT Was Trained
The NYT is asking a court in New York to proceed with their lawsuit, disputing all of OpenAI's arguments for dismissal. This lawsuit is one of several facing OpenAI over copyright issues. The outcome could significantly affect how AI models are trained in the future, potentially leading to new models based on licensed content.
Source: NYT to OpenAI: No hacking here, just ChatGPT bypassing paywalls (Ars Technica)
Image: sergeitokmakov from Pixabay
Comments
Post a Comment