Use of AI Waives Legal Protections

Rapid adoption of AI tools is reshaping many traditional corporate roles, but treating AI as a substitute for legal counsel can carry serious and unintended consequences. In a recent ruling of nationwide first impression, U.S. District Judge Jed Rakoff held that when a user communicated with a publicly available AI platform in connection with a legal matter (in that instance, a pending criminal fraud investigation), the AI user's communications were not protected by attorney-client privilege or the work product doctrine and had to be produced to the Government. See United States v. Heppner, Case 1:25-cr-00503-JSR (S.D.N.Y. 2/17/2026).

The legal dispute underlying this very consequential ruling arose from various corporate securities and wire fraud allegations for which the defendant (one Bradley Heppner) was indicted. In connection with his arrest, FBI agents seized documents that memorialized Heppner's communications with Anthropic's AI platform, "Claude," after he had been served with a grand jury subpoena and learned that he was the target of a Government investigation. Heppner used Claude (without direction or supervision from counsel) to outline a defense strategy in anticipation of his potential indictment. In resisting the Government's demand to inspect those documents, his counsel argued that Heppner had (1) input information into Claude that he learned from counsel; (2) created the AI Documents for the purpose of obtaining legal advice from counsel; and (3) subsequently shared the contents of the AI Documents with counsel. See id. at 3-4.

Noting the well-established doctrine that "attorney-client privilege attaches to, and protects from disclosure, 'communications (1) between a client and his or her attorney (2) that are intended to be, and in fact were, kept confidential (3) for the purpose of obtaining or providing legal advice'" (citation omitted), the Court concluded that Heppner's "AI Documents lack at least two, if not all three, elements of the attorney-client privilege," id. at 4-5, pointedly explaining, "[b]ecause Claude is not an attorney… that alone disposes of Heppner's claim of privilege." Id. Beyond the question of whether the communications were privileged, the Court further held that the communications memorialized in the AI Documents were not even confidential. The Court pointed to Claude's privacy policy, which permitted Anthropic not only to collect conversational inputs and outputs between the user and Claude, and to use such data to train Anthropic models, but also authorized Anthropic "to disclose such data to a host of 'third parties,' including 'governmental regulatory authorities.'" Id. at 6. Importantly, the Court noted that AI users do not have substantial privacy interests in their conversations with a publicly accessible AI platform "which the platform 'retains in the normal course of its business.'" Id. at 6-7 (citation omitted). Whether Heppner's communications with Claude were for the purpose of obtaining legal advice was a closer call, but "what matters for the attorney-client privilege is whether Heppner intended to obtain legal advice from Claude, not whether he later shared Claude's outputs with counsel. And Claude disclaims providing legal advice." Id. at 7-8 (emphasis in original).

In sum, "the communications between Heppner and Claude were not privileged at the time they took place," and "even assuming that Heppner intended to share these communications with his counsel and eventually did so, it is black-letter law that non-privileged communications are not somehow alchemically changed into privileged ones upon being shared with counsel." Id. Further, "even if certain information that Heppner input into Claude was privileged, he waived the privilege by sharing that information with Claude and Anthropic, just as if he had shared it with any other third party." Id. n. 3.

In parallel analysis, the Court noted that the work product doctrine "provides qualified protection for materials prepared by or at the behest of counsel in anticipation of litigation or for trial," id. at 9 (emphasis added, citation omitted), but is not intended to "shield[] from discovery materials in an attorney's possession that were prepared neither by the attorney nor his agents." Id. (citation omitted). The Court likewise concluded that "[t]he AI Documents do not merit protection under the work product doctrine because, even assuming, arguendo, that they were prepared 'in anticipation of litigation,' … they were nevertheless not 'prepared by or at the behest of counsel,' … nor did they reflect defense counsel's strategy." Id. (citations omitted).

In conclusion, the Court observed that "[g]enerative artificial intelligence presents a new frontier in the ongoing dialogue between technology and the law. Time will tell whether, as in the case of other technological advances, generative artificial intelligence will fulfill its promise to revolutionize the way we process information. But AI's novelty does not mean that its use is not subject to longstanding legal principles."

What Lessons Can Be Learned From Heppner?

• AI chatbots and agents are not attorneys and you should not assume that your communication with them (including about sensitive legal matters) are privileged or work product.^[1]
  • Accordingly, if you are involved in a dispute related to such communications, they may very well be discoverable by the other side.
  • To understand the implications of your use of and interactions with AI tools, you need to understand their applicable Terms of Use or Terms of Service, which may vary according to which version of a model you use. Moreover, as many AI tools pass data from your input to a foundational model like Claude, you must understand and appreciate how each vendor interacts with these models, and how inputs may be transmitted across interconnected vendors and foundational model providers, potentially undermining confidentiality and privilege.
• Developing and adhering to an appropriate AI policy is a basic precaution applicable to virtually every enterprise.
  • Control of what AI tools/applications are used and what data is exposed to them are critical elements of effective AI policies.
  • The deployment of AI technology is becoming ubiquitous, and understanding when and how you are interacting with AI can help you to manage associated risks.
• Keeping your "lawyer in the loop" can help to mitigate certain of the risks arising from AI use.
  • The substance of responses from AI chatbots and agents can be incomplete, misleading or false and should be vetted by someone with appropriate expertise.
• At present, there are very few legal or regulatory requirements for or constraints on AI technology (particularly in the U.S.), so parties are left to rely on self-help measures - technical, operational and contractual.
  • In the absence of law and regulation, managing risks and liabilities associated with AI through appropriate contract terms (i.e., with customers, vendors and service providers) represents one of the biggest levers that businesses can use to protect themselves, their interests and digital assets in the nascent Age of AI.
• To quote Judge Rakoff in Heppner: "AI's novelty does not mean that its use is not subject to longstanding legal principles."

^[1] Notably, OpenAI was recently sued by an insurer who alleged that the operator of the popular ChatGPT platform is practicing law without a license by providing legal advice to a long-term disability recipient who, in the absence of counsel and through the help of ChatGPT, had initiated a lawsuit against the insurer and rapidly filed more than 40 motions with the assistance of the artificial intelligence tool. See Nippon Life Ins. Co. of Am. v. OpenAI Found., No. 1:26-cv-02448 (N.D. Ill. filed Mar. 4, 2026).

Artificial Intelligence is reshaping industries at a rapid pace - introducing a wave of complex and novel legal challenges. From AI compliance and governance to intellectual property and risk management, businesses need trusted advisors who understand both this technology and the law.

Why Calfee?

As a full-service law firm with one of the largest and most experienced Intellectual Property and Information Technology practices in the Midwest, Calfee is uniquely positioned to deliver industry-leading counsel in this rapidly evolving space. Calfee's multidisciplinary approach combines IP, data privacy, cybersecurity, and regulatory expertise to deliver comprehensive solutions for businesses at the forefront of AI Innovation.

Calfee's Capabilities

Calfee's team of over 40 IP and IT attorneys leverages their strong technical expertise to deliver strategic, forward-thinking counsel and help clients navigate all legal aspects of AI so they can identify and maximize opportunities in this emerging technology.

Calfee, Halter & Griswold LLP is a full-service corporate law firm with 160 attorneys and professionals and offices in Cleveland, Columbus, Cincinnati, and Indianapolis. As a founding member of Lex Mundi, Calfee also offers international representation through a network of independent law firms with 22,000 attorneys in more than 125 countries.

Since 1903, Calfee's mission has been to provide meaningful legal and business counsel on matters critical to our clients' success. Calfee lawyers routinely represent a wide spectrum of private and public organizations - from emerging companies to Fortune 500 corporations - as well as government entities, nonprofit organizations, trade associations, and individuals.

Calfee is consistently ranked as one of the top law firms in Ohio and continues to receive recognition, both nationally and regionally, from a number of leading industry publications.