Court warns against chatbot legal advice in Heppner case

A federal court found that chats with a publicly available generative Artificial Intelligence tool were not protected by attorney-client privilege or the work-product doctrine. The ruling highlights litigation risks when executives or employees use chatbots for legal guidance without lawyer supervision.

A federal court recently ruled that Bradley Heppner’s chats with the Artificial Intelligence tool Claude were not protected by attorney-client privilege or the work-product doctrine. That meant the federal government could access his chatbot prompts, uploads, and responses, including material that could reveal whether Heppner knew his conduct was illegal. Heppner, who was charged with securities fraud and other federal crimes, had used Claude to draft and analyze documents related to his case and had also provided information he received from his lawyers.

The court concluded that the chatbot exchanges were not confidential, were not communications with counsel, and were not created for the purpose of obtaining legal advice. As a result, prosecutors could review and use them in the case. The ruling was described as fact dependent and, in its narrowest form, applied to conversations with a publicly available, non-enterprise, generative Artificial Intelligence platform that were not made at the request of counsel. United States v. Heppner, 2026 U.S. Dist. LEXIS 32697 Feb 17, 2026.

The decision underscores that conversations with a chatbot are generally not confidential under the terms of the most popular Artificial Intelligence models and may be discoverable in litigation. Feeding an attorney’s recommendations into a chatbot to obtain a “second opinion” can also destroy the privilege that would otherwise protect those attorney communications. The same risk extends beyond criminal matters to civil disputes, including employment claims and business litigation.

Businesses also face exposure when non-lawyer employees use Artificial Intelligence tools during workplace investigations, while handling employment complaints, or in preparation for litigation. Those uses may create records that adversaries can later obtain and use in court. The court indicated that the outcome might have been different if Heppner’s lawyers had supervised the chatbot use. The broader message is that Artificial Intelligence tools may be useful in business operations, but using them for sensitive legal questions without legal guidance can create significant and unnecessary risk.

68

Impact Score

Google Vids opens free video generation to all Google users

Google has made Google Vids available to anyone with a Google account, adding free access to video generation with its latest models. The move expands Google’s end-to-end video workflow and increases pressure on rivals that charge for similar tools.

Newsom orders California to weigh Artificial Intelligence harms in contract rules

Gov. Gavin Newsom has signed an executive order directing California agencies to account for potential Artificial Intelligence harms in state contracting while expanding approved use of generative tools across government. The move follows a dispute involving Anthropic and reflects a broader split between California and the Trump administration on Artificial Intelligence oversight.

Google launches Gemma 4 open model family

Google has introduced Gemma 4, a new family of open-weight Artificial Intelligence models focused on advanced reasoning and multimodal capabilities. The release expands the Gemma line with broader deployment options, stronger performance claims and a more permissive open source license.

PrismML launches 1-bit large language model family

PrismML has emerged from stealth with a $16.25 million seed round and an open source release of its 1-bit Bonsai large language models. The startup says the models sharply cut memory use and energy consumption while aiming to preserve performance on standard benchmarks.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.