Quarter 4, 2025: Can Your Chat GPT Logs Become Evidence in a Lawsuit?
By: Attorney Paige Hulse
Generative AI, like ChatGPT, has quickly become part of everyday business operations, providing a competitive business advantage. Courts are now grappling with how AI fits within existing legal frameworks, but one point is clear: ChatGPT prompts and outputs may be discoverable and may also create copyright exposure.
A recent ruling has illuminated the murky waters of AI legality. In late 2025, a federal magistrate judge in the S.D. New York ordered OpenAI to produce ~20M anonymized ChatGPT chat logs to The New York Times and other publisher-plaintiffs. The plaintiffs alleged OpenAI’s models were trained on copyrighted news articles without authorization. The Court rejected OpenAI’s argument that user privacy concerns should limit discovery because anonymization sufficiently addressed those concerns and that the logs were relevant to determining whether copyrighted material was reproduced, paraphrased, or otherwise relied upon by the model. OpenAI appealed the order, but the message was unmistakable: AI chat logs can be discoverable electronic evidence. This ruling has broader implications for businesses using AI. Courts are beginning to treat AI prompts and outputs as electronically stored information (ESI), which may be subject to preservation obligations and production in discovery. In some instances, courts have even required AI providers to preserve logs that would otherwise be deleted once litigation is reasonably anticipated.
Beyond discoverability, how AI is used matters, especially when prompts involve copyrighted material. Directly pasting another person’s protected work into a prompt, or instructing AI to “rewrite” or “mirror” a specific author’s or competitor’s content can create evidence of access and intent in a copyright infringement claim. While many debate whether or not the creativity behind inputs or ChatGPT training models constitute the legal defense of “fair use,” courts have not been favorable in such findings. Fair use is a fact-specific analysis, weighing factors such as purpose, transformation, amount used, and market impact. In short, AI-assisted use may still infringe if it closely tracks the original work, substitutes for it, or affects its market – particularly in commercial settings When used for business purposes, your ChatGPT interactions should be treated like business records with no assumption of privacy. Prompts that reference third-party works, proprietary content, or creative materials may later be scrutinized to determine whether infringement occurred or whether fair use applies. Best practices for use of AI in business should include refraining from pasting copyrighted or proprietary third-party content directly into AI prompts (or “inputs”) for a commercial purpose unless you have a license to do so. Additionally, operate under the assumption that AI logs may need to be preserved and produced in litigation even if deleted from your “chat history.”
AI is transformative regarding how creative and business work is performed, yet Court guidance accentuates that efficiency does not nullify copyright laws or discovery obligations. We encourage consulting with counsel before incorporating AI into sensitive or proprietary work. For trusted strategic guidance on navigating this developing area of law, contact Paige Hulse of the firm at 918-494-6868.