Date:
February 16, 2026
Author:
Ian R. Cohen
/
Principal
The Problem
A court recently ruled that a defendant who used a non-enterprise version of Claude did not qualify for attorney-client privilege for those chats: even though he was researching informaiton his legal counsel shared and sent it back to his attorneys for further discussion (United States v. Heppner (SDNY, Feb 10, 2026)).
The judges ruling focused on:
Lack of Counsel Oversight: Those LLM AI conversations being done outside the presence of counsel and not at the direction of counsel
Terms of Service: Claude being unable to provide legal advice as expressly stated in its terms
Third-Party Disclosure: Claude's terms also state prompts may be disclosed to regulatory authorities: there is no expectation of privacy or confidentiality
The Bottom Line: Prompting an AI is legally equivalent to a Google search or a public social media post. Sending those chats to your attorney after the fact will also not create attorney-client privilege. This is a major risk for those who aren't thinking about it or changing their actions today.
What it Means For You
This ruling serves as a vital reminder that everything you prompt in a consumer-grade AI is likely discoverable in a dispute or litigation later. AI might give you the tools to help you think like a lawyer and analyze problems, but it is still not a substitute for actual legal advice from a licensed attorney. The good news is there are cost effective way to protect yourself better without the exhorbitant expense of an hourly attorney or traditional law firm.
The Solution
Based on this ruling, I'm already working with my clients on the following:
AI-Specific Engagement Letters: Updated engagement letters that extend attorney-client privilege to AI to the greatest extent permitted by law
Low-Cost "Prompt Retainers": Implementing low cost retainers to establish an attorney-client relationship where we can colloaborate on prompts and protect AI output
Enterprise LLM Deployment: Utilizing enterprise level LLMs that have stricter data standards and provide greater levels of confidentiality and data priacy
Don't let your research become your opponents evidence. Contact IRC Legal today for a strategy session to learn more about AI guardrails that could keep you out of trouble.














