DHS Asks OpenAI to Unmask User Behind ChatGPT Prompts, Possibly First Such Case (gizmodo.com)

A unit of the Department of Homeland Security that investigates child sexual exploitation has asked OpenAI for records tied to a user it alleges ran a child-abuse website — a request Forbes calls the first known federal search warrant seeking ChatGPT user data. The reportedly unsealed court filings in Maine show the suspect discussed using ChatGPT with an undercover agent, but the actual prompts recovered by investigators were largely innocuous (examples cited include “What would happen if Sherlock Holmes met Q from Star Trek?” and an AI-generated, Trump‑style humorous poem excerpt). The underlying search-warrant affidavit is now sealed, though the criminal complaint — which cites context clues gathered in undercover chats (military service, locations lived and visited, a favorite restaurant, and weight/fitness details) — remains public and helped investigators narrow the suspect to a specific individual. The case is significant because it appears to mark the legal frontier for compelling AI companies to hand over conversational logs and related records, much as law enforcement already does with social platforms and search engines. Technically, it highlights two investigative realities: (1) AI systems retain conversation content and metadata that can be forensically valuable even when prompts seem unrelated to criminal activity, and (2) investigators can combine AI logs with external intelligence to identify users. The move raises precedent-setting questions about warrants, data-retention practices at AI providers, and privacy risks for users of chatbots as law enforcement increasingly treats them like other digital evidence troves.