A recent ruling by Judge Jed S. Rakoff in the Southern District of New York confirmed that documents generated using public AI tools, such as Anthropic’s Claude, are not protected by attorney-client privilege or the work-product doctrine. The court emphasized that AI cannot form a privileged relationship because it is not a licensed attorney, and users cannot reasonably expect confidentiality since public AI providers can retain and use the data. Additionally, the AI explicitly disclaims providing legal advice, and materials generated independently of counsel do not qualify as work product. This decision highlights that feeding privileged information into public AI platforms may waive confidentiality, and simply transferring AI-generated outputs to a lawyer does not retroactively make them privileged. Enterprise AI tools with contractual confidentiality protections may be treated differently, but the law on that remains unsettled.