AI and Attorney–Client Privilege: What Business Clients Need to Know After United States v. Heppner
Artificial intelligence tools are rapidly becoming part of everyday business operations. Managers use them to summarize documents, organize facts, and test arguments before speaking with counsel.
A recent ruling from the U.S. District Court for the Southern District of New York shows that this convenience can come with real legal risk.
On February 10, 2026, in United States v. Heppner, Judge Jed Rakoff ruled from the bench that materials a defendant created using Anthropic’s Claude and later provided to his lawyers were not protected by attorney–client privilege or the work product doctrine.
Even without a written opinion yet, the signal to businesses is straightforward: careless use of consumer AI platforms can undermine confidentiality.
What Happened
After receiving a subpoena and retaining counsel, the defendant independently used Claude to prepare written analyses describing potential defenses and legal theories. When federal investigators later obtained those files, defense counsel argued they were privileged because the documents were meant to help the client communicate with his attorneys.
The court rejected that position.
Attorney–Client Privilege
Attorney–client privilege protects communications between lawyer and client made for the purpose of seeking or providing legal advice and maintained in confidence.
Judge Rakoff held that confidentiality was lost once the information was shared with a third-party AI provider whose terms disclaimed any expectation of privacy. Sending the material to a lawyer afterward did not retroactively create privilege.
Work Product Doctrine
The work product doctrine typically protects materials prepared by or at the direction of counsel in anticipation of litigation.
Because the defendant generated the AI analyses on his own initiative—and not because a lawyer instructed him to do so—the protection did not apply.
Why This Matters for Businesses
Many executives assume that if a document is created to help a lawyer, it must be protected.
That assumption can be wrong.
If company personnel input sensitive facts, mental impressions, or strategy into an external system, a court may view that step as disclosure to a third party. The result may be that regulators or adversaries can later obtain the material in discovery.
In short, informal AI use can unintentionally create evidence.
Consumer vs. Enterprise Systems
The ruling focused on a public-facing platform. Enterprise systems often promise stronger contractual confidentiality protections and may represent that inputs are not used for model training.
While this does not automatically guarantee privilege, it can place a company in a stronger position to argue confidentiality was preserved.
Practical Guidance
Businesses can continue benefiting from AI, but legal coordination is essential.
Involve counsel early.
If a dispute or investigation is pending or reasonably anticipated, consult your attorney before using AI to analyze facts or strategy.
Make direction explicit.
If work is done to assist counsel, the record should reflect that it occurred at counsel’s request.
Use controlled environments.
Prefer platforms with negotiated confidentiality commitments over open consumer tools.
Keep your lawyers informed.
Counsel should understand what systems were used and for what purpose.
Document the legal objective.
Clear documentation makes privilege positions easier to defend.
The Broader Shift
Courts are applying familiar waiver principles to new technology. What has changed is how easily information can be transmitted outside the organization, often by well-meaning employees trying to be efficient.
AI governance is now part of litigation readiness.
How Good Pine P.C. Helps
Good Pine P.C. advises companies in New York and New Jersey on building practical structures that allow innovation while protecting legal rights. We work with businesses to:
develop internal AI-use policies,
create approval pathways when legal issues are implicated,
reduce inadvertent waiver risk, and
prepare defensible positions if discovery disputes arise.
AI can create efficiency. Without guardrails, it can create exposure.
Disclaimer
This publication is for general informational purposes only and is not legal advice. Reading it or contacting Good Pine P.C. does not create an attorney–client relationship. Legal outcomes depend on specific facts and applicable law. Prior results do not guarantee future outcomes. In some jurisdictions, this material may be considered attorney advertising.
