On Dec. 22, 2025, the New York City Bar issued a formal opinion on the ethics for AI use in the recording, transcription, and summarization of conversations between attorneys and clients. The opinion of the Professional Ethics Committee addressed the issues that may arise if the attorney or the client is the one employing AI programs, with an emphasis on notice.
The Attorney
The Bar likened the use of AI tools to other forms of recording that were already available to attorneys. Generally, even before this opinion, attorneys were discouraged from recording conversations with a client. AI has exacerbated many of the issues that formed the foundation of existing formal opinions. Those opinions were based upon technology that simply captured and preserved communications. AI takes this technology further to the extent that it draws inferences from the audio or visual data depending on what type of AI model is in use.
If an attorney is using AI to record, transcribe, or summarize a conversation with a client, there is no specific rule that addresses the responsibilities of an attorney. Instead, the New York Bar places that scenario under the general ethical prohibition against recording conversations without consent. The principal issue is not the recording itself, but rather the deceptive practice of recording the client without warning them first. Thus, to address those issues, an attorney should seek to fully inform the client prior to using AI tools to record the conversation. AI-generated summaries require additional consideration. Many states have laws regarding biometric data which includes voiceprints, and the Bar advises that attorneys must comply with such laws in using AI tools as the production of summaries often requires AI to store the biometrics of the client for some amount of time.
To comply with rules governing confidentiality, an attorney is advised by the Bar to maintain cognizance of where the data will be stored by the AI, for how long, and how to retrieve the data for discovery purposes. Additionally, how the AI training is conducted and what privacy safeguards are in place on the program are very relevant to preserving confidentiality and privilege. Many of these aspects are in the domain of the provider rather than the attorney, but that does not absolve the attorney of their duty to be informed on how the information is being handled.
The duty of competence presents an additional hurdle. The New York Bar under Rule 1.1 advises attorneys to independently review any summaries, transcripts, or analyses for accuracy. Furthermore, this duty requires attorneys to understand the tools on a less-than-technical level including the security and privacy features. The duty extends to the training and supervision of staff and subordinates.
The Client
The New York Bar summarized their analysis by acknowledging that a client’s actions may make it difficult for the attorney to fulfill their ethical obligations. Depending on the circumstances, it is proper for an attorney to ask that their conversations not be recorded, to include provisions in the intake agreements with clients that place restrictions on the use of AI, or to advise the client on the risks implicit in the use of such tools.
In both circumstances of AI usage, the Bar stressed that the attorney should keep confidentiality at the forefront. Ultimately, it is the responsibility of the attorney to maintain knowledge about the technology they employ in their practice. Client consent goes a long way towards addressing those ethical considerations, but the attorney needs to have a greater level of understanding of how AI works and what happens to the information to be truly competent under their Rule 1.1 obligations.
This article originally appeared on Freeman Mathis & Gary LLP. www.fmglaw.com
About the Authors:
Alex Diaz is a partner at Freeman Mathis & Gary LLP. alex.diaz@fmglaw.com
John M. Forbes is an associate at Freeman Mathis & Gary LLP. john.forbes@fmglaw.com