New York bar association says lawyers cannot bill for AI-saved time

For lawyers billing by the hour, the guidelines draw a line between work done with AI and time saved because of it. The association says lawyers may charge for time actually spent prompting an AI tool, reviewing or verifying its output, and revising it. They may not bill clients for the hours the technology saved.[3] That distinction makes the lawyer’s actual work on the matter central to the billing decision. The guidance also says fees should be reasonable and specific charges should be disclosed to clients.[2]

General training on how to use AI is not billable, according to the guidance.[2][3] The association distinguishes that from reasonable adaptation or application of AI for a particular matter, which may be charged to the client.[3] Lawyers assessing a time entry should therefore consider whether the work was general learning or was tied to a client’s specific case. The guidelines do not make every AI-related task billable: time spent must fit the stated distinction, and fees remain subject to the requirement that they be reasonable.[3][2]

The association classifies the general cost of acquiring and maintaining AI tools as law-firm overhead, rather than a client expense.[3] AI-tool costs may not be billed to a client unless the expense is for a service specially incurred for that case.[2] The guidelines also say a charge must be tied to a specific client matter and its basis disclosed.[4] For billing and finance teams, the practical distinction is between a firm’s general investment in its tools and an expense specifically incurred for a client’s case.

The guidelines say lawyers generally do not have to tell clients when AI is used for administrative tasks such as word processing, spelling and grammar checks, calendaring or timekeeping.[4] They cover disclosure of AI use in court submissions, which the report says many courts already require.[2] The association urges lawyers to check whether outputs reflect bias from training data, design or deployment, and warns that biased outputs could reproduce systemic inequities.[2] It also says lawyers remain responsible for supervising nonlawyers and vendors who use AI tools.[2]