𝗧𝗵𝗲 𝗙𝗮𝘀𝘁𝗲𝘀𝘁 𝗪𝗮𝘆 𝘁𝗼 𝗟𝗼𝘀𝗲 𝗮 𝗟𝗮𝘄𝘀𝘂𝗶𝘁 𝗬𝗼𝘂 𝗛𝗮𝘃𝗲𝗻’𝘁 𝗙𝗶𝗹𝗲𝗱 𝗬𝗲𝘁 Want to guarantee your opponent knows your legal strategy? Type it into an AI chatbot. Bradley Heppner was chairman of a publicly traded company under federal investigation. He hired Quinn Emanuel. Then, on his own, he used Claude to prepare 31 documents with legal arguments for his defense. Reasonable instinct. Organize your thinking. Help your lawyers. Except a federal judge just ruled none of it was privileged. Anthropic’s terms allow data collection, model training, and disclosure to third parties. No reasonable expectation of privacy. Everything Heppner typed is now available to prosecutors. First order: “AI helps me prepare my case faster.” Second order: “I just handed my playbook to the other side.” 𝗕𝘂𝘁 𝘁𝗵𝗲 𝗿𝗲𝗮𝗹 𝘀𝘁𝗼𝗿𝘆 𝗶𝘀 𝗺𝗼𝗿𝗲 𝗻𝘂𝗮𝗻𝗰𝗲𝗱. Quinn Emanuel admitted Heppner prepared the documents on his own. They never directed him to do it. That concession killed the work product argument more than the AI angle did. Why? Because the government’s own brief acknowledged a scenario where AI-generated work product could be protected, if counsel had directed the searches. Different instruction from his lawyers, very different headline. 𝗡𝗼𝘁 𝗮𝗹𝗹 𝗔𝗜 𝗶𝘀 𝗰𝗿𝗲𝗮𝘁𝗲𝗱 𝗲𝗾𝘂𝗮𝗹. Heppner used a $20/month consumer account. Enterprise AI accounts contractually prohibit training on customer data and include confidentiality protections. No court has tested this yet, but legal commentary from firms like Debevoise suggests enterprise tools substantially strengthen the privilege argument. 𝗪𝗵𝘆 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗼𝘄𝗻𝗲𝗿𝘀 𝘀𝗵𝗼𝘂𝗹𝗱 𝗰𝗮𝗿𝗲. This isn’t limited to criminal cases. Civil litigation, employment disputes, partnership disagreements, insurance claims. Every legal question you type into a consumer AI tool is potentially sitting in a database, waiting to be subpoenaed. The fix isn’t to stop using AI. If you’re in litigation or anticipate it, let your lawyer direct the AI work. In writing. Use enterprise-tier AI for legal matters. Or let your lawyer use their own tools, where privilege attaches to their work. The question isn’t whether AI is useful for legal thinking. It’s whether your opponent would thank you for using it. Full case details: https://lnkd.in/esJ8w5s6