Chatbot chats discoverable

A federal judge in New York ruled that chats with an artificial intelligence chatbot can be used in court, not hidden behind attorney-client privilege. (reuters.com) The ruling came in *United States v. Heppner*, a securities-fraud case in the Southern District of New York. Judge Jed Rakoff said on February 10, 2026, and in a written memorandum filed February 17, that 31 documents Bradley Heppner created with Anthropic’s Claude were not privileged. (courtlistener.com) (debevoise.com) Rakoff wrote that no attorney-client relationship exists between a user and a public chatbot, and he said any privileged facts typed into Claude were waived by sharing them with Anthropic. He also rejected work-product protection for the chatbot-generated material. (mindingyourbusinesslitigation.com) (harvardlawreview.org) Attorney-client privilege is the rule that usually keeps private communications between a lawyer and client for legal advice. Public chatbots work more like a third-party service provider than a law office, and that distinction is now showing up in court orders and client alerts. (greenberglaw.com) (foxrothschild.com) Law firms have started rewriting their warnings to clients. Reuters reported that some firms now tell clients not to paste legal advice, case strategy, medical records, or internal company facts into consumer tools such as ChatGPT or Claude unless a lawyer directs the use and confidentiality terms are in place. (reuters.com) (shertremonte.com) The warning reaches beyond criminal cases. Lawyers writing about the decision said civil litigants and companies should assume prompts and outputs from public artificial intelligence tools may be discoverable in lawsuits, investigations, and regulatory reviews. (orrick.com) (crowell.com) At the same time, Apple has been pressing on a different risk: what these apps generate in public. NBC News reported on April 15 that Apple told senators it had warned Elon Musk’s xAI in January that Grok could be removed from the App Store over sexualized deepfakes if the company did not fix the app. (nbcnews.com) According to that letter, Apple found both Grok and X in violation of App Review Guidelines, rejected an early fix, and later approved changes that kept the app in the store. CNET and PCMag, citing the NBC report, said the dispute centered on nonconsensual sexualized imagery generated through Grok. (cnet.com) (pcmag.com) xAI did not immediately comment in the reports, and Apple’s letter said the company had worked with the developer to bring the app into compliance rather than remove it. In court and in app stores, the same point is getting sharper: chatbot records are being treated less like private thoughts and more like ordinary digital evidence. (nbcnews.com) (reuters.com)