The NYT is not giving up in the battle with OpenAI for copyright content.
The New York Times responded to OpenAI's claims that the publication allegedly used ChatGPT hacking methods to prepare a lawsuit against OpenAI. The publication categorically rejects OpenAI's accusations, describing them as untrue and untrue.
In the lawsuit, OpenAI claimed that the NYT had made "tens of thousands of attempts" to force ChatGPT to reproduce snippets of NYT articles. The company accuses the newspaper of using deceptive queries, in particular, consecutive questions like "what is the next offer?". According to OpenAI, this behavior is aimed at identifying two rare and unintended phenomena: repeated use of training data and model hallucinations, which the company considers an error and plans to fix.
However, the NYT defends its query methods aimed at identifying the use of stored data, including more than 100 articles of the publication. The newspaper points out cases when users often use ChatGPT to generate complete articles in order to bypass subscriptions.
The document also states that the NYT does not have information about how many and which articles were used in training GPT-3 models and subsequent OpenAI models, since the company did not disclose the composition of the training datasets of its models. The newspaper argues that its actions were aimed at identifying copyright violations, and not at creating grounds for a lawsuit.
The document also mentions the introduction of the Bing search feature for ChatGPT in May 2023, which allowed content to be obtained outside the model's training dataset, violating copyright by displaying synthetic search results that paraphrase NYT articles obtained and copied in real time by user requests. The option was temporarily disabled by OpenAI in July.
Before implementing the search function, the NYT contacted OpenAI with a notification that the company's tools violate the publication's copyright, but OpenAI ignored the NYT's claims.
The NYT insists that the court reject OpenAI's motion to dismiss the case, emphasizing that all the company's arguments failed. The newspaper emphasizes that this case is radically different from previous lawsuits filed by the authors of the books, and requires not only damages, but also a permanent injunction against violations by ChatGPT.
The NYT is seeking not only compensation for damages due to alleged copyright infringement, but also an injunction that could force OpenAI to start working on the AI model anew. New work can possibly occur on the basis of licensed content, so that publishers are always paid for using their data in training.
The New York Times responded to OpenAI's claims that the publication allegedly used ChatGPT hacking methods to prepare a lawsuit against OpenAI. The publication categorically rejects OpenAI's accusations, describing them as untrue and untrue.
In the lawsuit, OpenAI claimed that the NYT had made "tens of thousands of attempts" to force ChatGPT to reproduce snippets of NYT articles. The company accuses the newspaper of using deceptive queries, in particular, consecutive questions like "what is the next offer?". According to OpenAI, this behavior is aimed at identifying two rare and unintended phenomena: repeated use of training data and model hallucinations, which the company considers an error and plans to fix.
However, the NYT defends its query methods aimed at identifying the use of stored data, including more than 100 articles of the publication. The newspaper points out cases when users often use ChatGPT to generate complete articles in order to bypass subscriptions.
The document also states that the NYT does not have information about how many and which articles were used in training GPT-3 models and subsequent OpenAI models, since the company did not disclose the composition of the training datasets of its models. The newspaper argues that its actions were aimed at identifying copyright violations, and not at creating grounds for a lawsuit.
The document also mentions the introduction of the Bing search feature for ChatGPT in May 2023, which allowed content to be obtained outside the model's training dataset, violating copyright by displaying synthetic search results that paraphrase NYT articles obtained and copied in real time by user requests. The option was temporarily disabled by OpenAI in July.
Before implementing the search function, the NYT contacted OpenAI with a notification that the company's tools violate the publication's copyright, but OpenAI ignored the NYT's claims.
The NYT insists that the court reject OpenAI's motion to dismiss the case, emphasizing that all the company's arguments failed. The newspaper emphasizes that this case is radically different from previous lawsuits filed by the authors of the books, and requires not only damages, but also a permanent injunction against violations by ChatGPT.
The NYT is seeking not only compensation for damages due to alleged copyright infringement, but also an injunction that could force OpenAI to start working on the AI model anew. New work can possibly occur on the basis of licensed content, so that publishers are always paid for using their data in training.
