Overview
- Two recent federal rulings reached opposite results on AI materials, with a Michigan court treating a pro se litigant’s ChatGPT drafting as protected work product and an SDNY court rejecting privilege and work-product claims for a defendant’s Claude chats.
- Judges focused on classic tests, weighing who used the tool, whether a lawyer directed the exchange, and if platform terms allowed the AI provider to collect or disclose prompts and outputs.
- In the SDNY case, the court said public platform terms undercut any reasonable expectation of secrecy and found no attorney direction, so inputs and outputs were fair game in discovery.
- Employment counsel now expect requests for AI prompts, outputs, auto-drafted reviews, and meeting transcripts, and the guidance urges vendor vetting, clear policies, training, tighter retention, and using secure, non-retaining enterprise AI under counsel’s direction.
- JD Supra frames the issue as an emerging split with practical steps for employers, while National Law Review warns that pasting privileged content into a public chatbot can waive protection and invite subpoenas and regulator questions about AI use.