OpenAI claims NYT manipulated ChatGPT to copy its articles
OpenAI has responded to a copyright lawsuit filed by the New York Times (NYT), calling it "without merit" and expressing hope for a future partnership. The NYT accuses OpenAI's ChatGPT of reproducing its articles, but OpenAI claims the media outlet manipulated prompts to make the AI tool copy excerpts. OpenAI also stated that using internet data to train AI models falls under fair use rules.
Allegations of verbatim reproduction
The NYT alleges that ChatGPT copied its stories word-for-word, which OpenAI denies. The AI company said, "Even when using such prompts, our models don't typically behave the way the NYT insinuates, which suggests they either instructed the model to regurgitate or cherry-picked their examples from many attempts." OpenAI admits to removing a ChatGPT feature called Browse that unintentionally copied content.
Fair use and access to copyrighted content
OpenAI argues that AI models need access to "the enormous aggregate of human knowledge" to learn and solve problems. The Microsoft-backed firm claims that using copyrighted works for training AI models is protected by fair use under copyright law. OpenAI recently made a similar argument to the UK House of Lords, stating that AI systems like ChatGPT can't be built without access to copyrighted content and must include such works to represent human intelligence and experience.
NYT's response and legal stance
The NYT disputes OpenAI's fair use claim, arguing that ChatGPT and Microsoft's Bing chatbot offer similar services as the newspaper. Ian Crosby, lead counsel for the NYT, said, "That's not fair use by any measure." The NYT claims that the AI tools sometimes provide exact copies of sections of its reports, which OpenAI says is a "rare bug" that it is "working to drive to zero."
OpenAI's partnerships and negotiations with publishers
OpenAI has partnered with Axel Springer and the Associated Press, compensating them for using their content to train its models. The NYT claims it spent months negotiating with OpenAI and Microsoft for a similar licensing agreement but couldn't reach a fair resolution. Despite the lawsuit, OpenAI remains hopeful for a constructive partnership with the publication. The company also lets publishers opt out of having their sites scraped by its tools. Per OpenAI, the NYT adopted this option in August 2023.
