by Joseph Brophy for the Maricopa Lawyer, a publication of the Maricopa County Bar Association
A federal court in New York recently made waves when it ruled that a client’s chats with artificial intelligence (AI) were not protected by the attorney-client or work product privileges, even if the information the client provided to AI came from the client’s attorney. In the same vein, the New York City Bar Association recently released Formal Opinion 2025-6, titled “Ethical Issues Affecting Use of AI to Record, Transcribe, and Summarize Conversations with Clients.” The Opinion addresses two distinct scenarios: (1) when an attorney uses an AI platform to record and summarize client calls or meetings; and (2) when the client uses an AI platform to record meetings with their attorney.
There are several AI applications that attorneys can use to transcribe conversations with a client, such as Otter, Fireflies and Descript. These applications replace old-fashioned attorney note taking. Using AI transcriptions theoretically ensures an accurate record of what was said and solves the problem of having to later decipher the written notes of a lawyer who writes like a third grader. When an attorney uses these applications, several ethical rules are implicated.
ER 8.4 (dishonesty) requires that the client knows and consents to the conversation being recorded. This is because most clients presume their privileged conversations with their attorney will not be recorded and clients are generally inclined to choose their words more carefully, or change what they say entirely, if they know they are being recorded. Moreover, although AI is new, recording technology is not. The overwhelming weight of authority regarding the audio recording of attorney-client communications by a lawyer provides that the fact of the recording must be disclosed to the client.
ER 1.1 (competency) requires a lawyer to ensure that the transcription of the conversation is accurate. Of particular concern are conversations where an attorney offers informal advice in the moment that may need to be changed based on further research. Significant concerns arise if inaccurate advice (even if unintentional) from the attorney is memorialized without being reviewed and corrected if necessary. AI tools may also create advice on their own, which similarly must be reviewed for accuracy before relying on it. By now lawyers should be aware of the perils of blind or uninformed reliance on AI-generated legal advice, arguments or authority.
Client confidentiality is also an ever-present concern with AI. Many AI tools are not clear whether they retain information input into their system. Some applications are quite clear that they reserve the right to disclose any information provided by a user to third parties. Disclosing otherwise confidential information or privileged client communications to a third-party for a transcription or summary may inadvertently violate confidentiality or waive the privilege. In fact, this was a critical point raised by the New York federal court in the Heppner decision mentioned above. The privacy policy of the AI used by the client in Heppner (Claude) was clear that it reserved the right to disclose information to third parties. That fact was critical to the court’s finding that no privilege applied or was waived.
An attorney must be aware of the data-retention and privacy policy of any communicative tool they use, including Zoom or Microsoft Teams. Any AI application not made specifically for lawyers (such as those offered by Westlaw or Lexis) should be presumed to retain any information provided to it and the right to provide that information to third parties.
Regarding a client’s use of AI, lawyers do not control what their clients do or to whom (or what) they provide information. Few clients appreciate the risks associated with AI in the context of a legal representation. To that end, the Opinion offers several points of guidance to attorneys.
First, the expectations regarding AI use should be relayed at the outset of the representation. This can be in the form of a letter to the client communicating the firm’s policies regarding AI, or such information can be included in the engagement agreement. Second, attorneys should include provisions in engagement agreements providing that any recordings, transcripts, or summaries prepared by AI tools selected or used by the client will not be deemed dispositive or binding as against the attorney unless they are promptly provided to the attorney so that they can conduct independent reviews of the accuracy of these materials. This prevents the attorney from being held to a statement that was inaccurately recorded or misstated by the client’s AI. Finally, attorneys should inform their clients about the dangers of AI use and the potential waiver of confidentiality and privilege in doing so.
There are few, if any, ethical opinions discussing lawyers’ use of AI recording or note taking applications. If you are among those who utilize those tools in your practice, the New York City Bar Association’s Formal Opinion 2025-6 is required reading unless and until Arizona issues its own guidance.
Download the article here.