ARTICLE
26 February 2026

Your AI-Generated Documents Are Not Protected According To First-Ever Federal Court Ruling

BL
Butzel Long

Contributor

Founded in 1854, Butzel Long has played a prominent role in the development and growth of several major industries. Business leaders have turned to us for innovative, highly-effective legal counsel for over 170 years. We have a long and successful history of developing new capabilities and deepening our experience for our clients’ benefit. We strive to be on the cutting edge of technology, manufacturing, e-commerce, biotechnology, intellectual property, and cross-border operations and transactions.

As businesses increasingly turn to artificial intelligence (AI), a recent federal court ruling now confirms for the first time that information shared with public AI platforms, even for information believed ...
United States Technology
Marcella Eid’s articles from Butzel Long are most popular:
  • in United States
Butzel Long are most popular:
  • within Accounting and Audit, Litigation, Mediation & Arbitration, Media, Telecoms, IT and Entertainment topic(s)
  • with readers working within the Technology industries

As businesses increasingly turn to artificial intelligence (AI), a recent federal court ruling now confirms for the first time that information shared with public AI platforms, even for information believed to be protected by attorney-client privilege or work-product doctrine, may later be discoverable and used as evidence. This ruling serves as a stark warning for all businesses and individuals that rely on public AI platforms for legal matters, especially where confidentiality and privilege are at risk.

Brief Background

In United States v. Heppner, Bradley Heppner used the AI platform Claude to prepare defense strategy documents after learning he was the target of a criminal investigation stemming from his alleged misconduct as an executive of several corporate entities.1 Heppner argued that the documents were protected by attorney-client privilege and the work product doctrine because the documents contained information he had learned from his attorney, were created to acquire legal advice from his attorney and were ultimately shared with that attorney.2 In the end, Judge Rakoff rejected Heppner's assertions of attorney-client privilege and work-product doctrine.

It is perhaps obvious, but important to emphasize in this new world of AI-generated communications and documents, that for a communication to be protected by attorney-client privilege, the communication must be between an attorney and the attorney's client. The privilege is lost if it is communicated outside of the parties sharing the attorney-client privilege. A third-party tool that does not protect the confidentiality of the communication would vitiate privilege. For documents to be protected under the work product doctrine, the “documents and tangible things” must be created by or at the direction of counsel or created “in anticipation of litigation or for trial.”3

Key Issues

The Court addressed these two key privilege issues as follows:

  1. Attorney-Client Privilege:  First, the Court explained that the AI platform was not an attorney, and thus the documents were not communications between Heppner and his counsel.4 Second, the documents lacked confidentiality due to the AI platform's privacy policy which allowed for data collection, training use, and third-party disclosure.5 Third, Heppner did not communicate with the AI platform for the purpose of obtaining legal advice, rather he communicated with the AI platform on his own volition and from an AI platform that explicitly disclaimed providing legal advice.6 Last, the Court provided a reminder that non-privileged materials already in existence do not become privileged when later shared with an attorney.7
  2. Work Product Doctrine: Heppner's attorney conceded that the documents were prepared by Heppner on his own volition.8 The Court found that such documents do not merit work product protection because they were not prepared "by or at the behest of counsel" and did not reflect his counsel's strategy.9

Key Takeaways and Implications

Although the ruling may not come as a surprise based on the current well‑established law on attorney‑client privilege and work product doctrine, it nevertheless signals potentially significant consequences for businesses and individuals that utilize public AI platforms for legal matters.

Remember:

  • Communications with public AI platforms may be discoverable.
  • Consult a licensed attorney, not a public AI platform, for legal or compliance advice.
  • Carefully examine AI platforms' privacy policies before use. Most if not all public AI platforms permit data training from users' input, and once the user's input is submitted to these public AI platforms, all privilege and confidentiality protections are lost.
  • Educate employees about privilege risks when using AI tools for work-related legal matters.

Footnotes

1. Memorandum as to Court's Ruling that Documents the Defendant Generated Through an Artificial Intelligence Tool Are Not Privileged, United States v. Heppner, No. 25-cr-00503-JSR (S.D.N.Y. Feb. 17, 2026).

2. Id.  at 3-4.

3. Federal Rule of Civil Procedure 26(b)(3)(A).

4. Heppner at 5.

5. Id. at 6.

6. Id. at 7-8.

7. Id. at 8.

8. Id. at 9-10.

9. Id. at 9.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.

[View Source]

Mondaq uses cookies on this website. By using our website you agree to our use of cookies as set out in our Privacy Policy.

Learn More