A major legal dispute between OpenAI and the New York Times has led to a significant shift in how the company handles user data. In December 2023, the Times sued OpenAI, alleging that the company used its copyrighted content to train its AI models without permission. The lawsuit sparked a sweeping court order requiring OpenAI to retain user chat logs—including those that had been deleted—“indefinitely” to preserve potential evidence for the case. The order, issued by U.S. Judge Ona Wang, drew widespread criticism for its broad scope and impact on user privacy. OpenAI argued that the requirement violated users’ expectations of privacy, especially since it affected hundreds of millions of ChatGPT users worldwide. Brad Lightcap, OpenAI’s COO, called the demand “sweeping and unnecessary,” describing it as a baseless attempt by the plaintiffs to force the company to retain vast amounts of user data. However, the situation has now changed. On Thursday, Judge Wang approved a joint motion submitted by both OpenAI and the New York Times, effectively lifting the indefinite retention order. This means OpenAI can now delete chat logs that users have removed, marking a significant victory for the company and a relief for users concerned about data privacy. That said, the settlement is not a complete end to data retention. According to Ars Technica, deleted and temporary chats will still be monitored for certain users, though it remains unclear which users or under what conditions. The chat logs that have already been preserved will remain accessible to the news organizations involved in the case, as part of ongoing efforts to investigate whether AI-generated outputs infringed on the Times’s content or falsely attributed information to its publications. While the immediate crisis over chat log retention has been resolved, the broader legal battle over AI and copyright remains unresolved. OpenAI has faced numerous lawsuits from media companies and content creators over the use of copyrighted material in AI training. Other AI firms, including Anthropic and Stability AI, are also involved in similar legal challenges. These cases are shaping the future of generative AI, as courts grapple with how copyright law applies to machine learning systems trained on vast amounts of public and licensed data.