OpenAI Accuses New York Times Of “Hacking” ChatGPT To Produce Copyrighted Materials
One of the largest newspapers in the world is accusing one of the most popular generative AI apps of producing copyrighted materials. The New York Times alleges that ChatGPT was trained on their newspaper articles and produces content that is similar or identical to copyrighted material. At present, ChatGPT has won most of the claims against them with judges deciding that the plaintiffs have failed to prove that ChatGPT causes any economic harm and that the output is not sufficiently similar to the original texts on which it was trained. However, the New York Times has produced new evidence that shows that OpenAI’s chatbot does produce content that violates its copyright.
OpenAI has countered the argument by claiming that the New York Times employed a prompt engineer who massaged ChatGPT into spitting out content that appeared to violate its copyright.
Understanding the the lawsuit
The New York Times is one of a number of plaintiffs that have sued OpenAI over the use of copyrighted material. Recently, a judge dismissed several complaints against the AI company on the grounds that the plaintiffs failed to show economic damage and that the material that ChatGPT produces is not sufficiently similar to any copyrighted material. One complaint survived, however, and this complaint alleges that it’s unlawful for ChatGPT to use copyrighted material without the express consent of the copyright owner.
In response, the New York Times has hired a prompt engineer to get ChatGPT to spit out material that is similar to New York Times articles. OpenAI is accusing the plaintiff of “hacking” ChatGPT to produce this material. They claim that the average user looking for information would not provide prompts that place ChatGPT into the position of spitting out copyrighted material.
AI companies have long argued that they have the right to use this material under Fair Use. Fair Use allows AI companies to use copyrighted materials to train its language models on copyrighted data. However, it cannot reproduce copyrighted materials verbatim. The New York Times is attempting to prove that ChatGPT produces articles that are nearly identical to their copyrighted materials. OpenAI claims that they have not been provided with any results that violate the New York Times’ copyright prior to them filing the lawsuit.
In a bid to prove that OpenAI is producing copyrighted material, the New York Times submitted evidence that the chatbot creates output that is near verbatim material from their copyrighted articles. If true, it could help the New York Times win on claims of copyright infringement. OpenAI is arguing that verbatim output is a “rare bug” and for the New York Times to come up with these examples, it must have “hacked” the chatbot by employing a prompt engineer who fed in thousands of prompts to solicit the copyrighted material.
Talk to a Copyright Infringement Attorney Today
If you’ve had your copyrighted materials misused by another party, talk to the Columbus, OH business torts attorneys at Kohl & Cook Law Firm, LLC. Our attorneys will provide you with a comprehensive strategy to get the infringing party to stop using your copyrighted material and recover damages related to its misuse.
Source:
cpomagazine.com/cyber-security/openais-defense-in-copyright-lawsuit-new-york-times-hacked-chatgpt-to-create-evidence/
