AI Misuse Nearly Misapplied Right to Counsel in Heppner Case
Judge ruled AI-generated documents were not protected by attorney-client privilege.
Why it matters: AI tools can misapply critical legal rights, risking confidentiality and defense strategies. Human oversight remains essential to catch such errors and safeguard constitutional protections.
- In February 2026, Judge Jed S. Rakoff ruled that documents created with AI tool Claude were not privileged.
- Defendant Bradley Heppner used Claude to generate 31 documents related to his securities fraud case.
- Court found AI tools like Claude do not create privileged attorney-client communications as they are not licensed attorneys.
- Heppner waived privilege because Claude’s privacy policy allowed data collection and third-party disclosure.
In a notable February 2026 decision, Judge Jed S. Rakoff of the U.S. District Court for the Southern District of New York addressed the legal status of documents generated by an AI platform. Bradley Heppner, charged with securities fraud, used Anthropic's Claude to create roughly 31 documents containing legal and factual queries related to his defense.
The court ruled these AI-generated documents were not protected by attorney-client privilege or the work product doctrine, as AI programs like Claude lack the necessary human legal relationship and are not licensed attorneys. As Judge Rakoff stated, "The use of such applications is not intrinsically privileged."
Furthermore, the ruling highlighted that because Claude's privacy policy permitted the collection and potential disclosure of user data, Heppner effectively waived any privilege. This exposes a critical risk: reliance on AI platforms without careful scrutiny may undermine confidentiality and legal protections.
This case underscores an important point for legal professionals deploying AI: while AI can assist with research and document generation, it cannot replace licensed attorneys, and its use may inadvertently compromise privileged information.
Similar concerns regarding AI errors recently emerged in law enforcement. In June 2026, a Florida man filed a lawsuit against Jacksonville Beach Police alleging wrongful arrest caused by an AI facial recognition error. Together, these examples reveal the ongoing challenges of integrating AI into critical legal and law enforcement contexts without robust human oversight to correct errors and protect individual rights.
By the numbers:
- 31 — documents generated by Heppner using Claude
- February 2026 — Judge Rakoff’s ruling on AI-generated documents
- June 2026 — Florida man’s lawsuit over AI facial recognition error