Quandaries & Quagmires: Ethical considerations in billing for AI-assisted work

Attorneys have a powerful new tool for automating research, drafting documents, and enhancing efficiency. Currently, many of the non-publicly available generative AI platforms like Thomson Reuters Co-Counsel or Bloomberg Draft Analyzer are pricey, which leaves law firms engaging in a cost-benefit analysis. This technological leap raises critical ethical questions: How should attorneys bill for AI-assisted work?

The answer lies in following well-established ethics rules on legal fees, which remain applicable despite technological advances. Reasonableness is the fundamental principle guiding attorneys’ billing practices in Minnesota and other jurisdictions. By balancing innovation with ethical standards, attorneys can leverage generative AI while maintaining compliance and ensuring fair fees and expenses. Now is the perfect time to review recent ethics guidance on attorney fees, which includes several state ethics opinions and guides offering specific advice on billing for generative AI-assisted legal work.

Attorney’s fees and expenses must be reasonable

Minnesota Rule 1.5(a) and (b) contain the basic ethical framework for billing clients. Rule 1.5(a) states that an attorney’s fee must be reasonable, considering factors like the time and labor required, the novelty and difficulty of the questions involved, and the skill needed to perform the service properly. Rule 1.5(b) emphasizes the duty of communication, requiring attorneys to clearly explain the basis or rate of the fee and expenses, preferably in writing, before or shortly after starting the representation.

Longstanding ethics guidance from the American Bar Association (ABA) remains relevant for billing AI-assisted legal work. ABA Formal Opinion 93-379 “Billing for Professional Fees, Disbursements and Other Expenses” (1993) emphasizes reasonable and transparent billing practices, warning against unreasonable fees such as duplicate billing, bill padding, and surcharges. ABA Formal Opinion 00-420 “Surcharge to Client for Use of a Contract Lawyer” (2000) discusses the ethical implications of outsourcing services and is analogous to contracting with a vendor for generative AI services.

Many states have issued ethics opinions or best practice guidance about using artificial intelligence, including California, Colorado, Florida, Maryland, Michigan, New Jersey, New York, Pennsylvania, Texas, and the District of Columbia. While the OLPR has not yet provided specific guidance on generative AI, insights from these states emphasize a consistent message: Attorneys must understand AI’s risks and benefits, ensuring its use aligns with the Rules of Professional Conduct. Key points include the reasonableness of fees, clear client communication, and transparent billing practices.

Billing factors that help ensure reasonableness

To ensure fees and expenses remain reasonable when using generative AI, attorneys should consider the following factors:

Time and labor: Generative AI can dramatically reduce the time needed for specific tasks, but duplicative or inflated fees are real risks, particularly with hourly billing arrangements, if AI substantially expedites the work. Florida Opinion 24-1 notes, “Though generative AI programs may make a lawyer’s work more efficient, this increase in efficiency must not result in falsely inflated claims of time.”¹  The District of Columbia bar cautions, “The same is true when the use of GAI reduces billable time, and the lawyer’s fee agreement with the client is based exclusively on the time the lawyer spends working on the matter. No matter how good or valuable the GAI’s output is, absent a different fee arrangement, the lawyer can only bill for the time the lawyer spent.”² The California State Bar advises that generative AI can be used to create efficiencies in work product. Lawyers can “charge for actual time spent” but “must not charge hourly fees for the time saved by using generative AI.” California guidance includes examples of permissible actual time spent, such as “crafting or refining generative AI inputs and prompts and editing generative AI outputs.”³

Time and labor considerations apply equally to billing entries attributed to nonlawyers under Rule 5.3 duties to supervise nonlawyer assistants. Florida notes that lawyers must supervise staff billing entries. For example, Florida uses the permissible example of a separate itemization for paralegal research, but not if the charges are accounted for in the attorney’s overhead. Alternatively, Florida advises attorneys to consider crediting the paralegal’s time against their own.

Skill and Expertise: Using generative AI requires expertise to ensure proper use and accurate interpretation. Technological competency is required of lawyers. See Cmt. 8 to Rule 1.1. Attorneys should account for their skill in using AI as a factor in determining fees but should ensure that the charges are not excessive relative to the actual work performed. Attorneys should refrain from billing time learning how to use a new technology. Florida notes that attorneys “should be careful not to charge for the time spent developing minimal competence in the use of generative AI.”⁴

Transparency: Consistent with Rule 1.5(b), attorneys must clearly communicate to clients how AI is used in their matter and how it affects the billing. Transparency builds trust and helps clients understand the value they receive. State ethics opinions uniformly advise informing clients about fees and costs associated with AI-assisted work. New York guidance provides, “If you will add a “surcharge” (i.e., an amount above actual cost) when using specific Tools, then you should clearly state such charges in your engagement letter, provided that the total charge remains reasonable.”⁵

Expenses: Charges should reflect the actual use and benefit to the client. Overestimating the value or usage of AI services to inflate costs is unethical under Rule 1.5(a). See also ABA Formal Opinion 93-379 (attorneys must not charge clients more than the actual cost incurred unless the client has agreed otherwise). Costs associated with generative AI are divided into two categories: direct costs and overhead costs. Direct costs, such as software subscriptions or licensing fees used explicitly for a client’s case, can be passed on to the client with full disclosure, client consent, and at actual cost. The D.C. bar advises, “If the lawyer intends to bill the client for use of GAI for which there is an out-of-pocket cost to the lawyer, that expected cost is an expense that should be communicated to the client under this rule.”⁶ Costs associated with general overhead, such as firm-wide AI tools used across multiple clients, should not be billed to individual clients as direct expenses. Instead, these costs are considered part of the firm’s operational expenses. Florida warns against prorating the costs of generative AI platforms. “If a lawyer is unable to determine the actual cost associated with a particular client’s matter, the lawyer may not ethically prorate periodic charges of the generative AI and instead should account for those charges as overhead.”⁷

Consider alternative fee arrangements (AFAs)

AFAs provide diverse ways for clients to pay their attorneys, focusing on outcomes and value rather than billable hours. Given AI’s efficiencies, there may be better or more ethical options than traditional hourly billing. Florida advises considering AFAs over hourly rates: “In the alternative, lawyers may want to consider adopting contingent fee arrangements or flat billing rates for specific services so that the benefits of increased efficiency accrue to the lawyer and the client alike.”⁸  Some AFAS include:

Flat fee: A set amount for specific services, regardless of time or effort spent. Generative AI’s efficiency aligns well with this model, ensuring cost predictability, and is well suited to transactional work.

Contingency billing: A risk-based model where clients pay a percentage of any monetary award if a successful outcome is obtained. Faster research and drafting produce better outcomes, benefiting clients and attorneys. Suitable for litigation work.

Success fee: Received upon achieving positive outcomes. Leveraging generative AI’s speed and accuracy enhances the chances of success and may offer opportunities for justifying success-based compensation.

Subscription fees: Recurring payments for ongoing services, including AI-assisted tasks, provide consistent revenue and service predictability for clients and attorneys.

Task-based billing: Breaking legal work into discrete tasks and assigning a set fee for each task aligns with the efficiencies gained from AI and ensures fair billing for each task.

Hybrid models: Due to their flexibility, hybrid models are gaining traction. One type of hybrid model that may work for AI-assisted legal work is combining a base flat fee with an additional success fee. Clients benefit from the flat fee by knowing the upfront cost of routine legal tasks, and the success fee motivates attorneys to leverage AI to achieve the best outcome efficiently.

 Clarity and transparency in fee agreement construction

When constructing a fee agreement with a client, it is crucial to recognize that one size does not fit all. Under Rule 1.5(b) and related ethical duties of communication, attorneys must clearly convey the rate and basis of their fees. Any ambiguities in these agreements are typically construed against the attorney as the drafter. One significant ethical risk is the potential lack of transparency in how AI tools are used and billed. The risk of clients misunderstanding how AI is used and billed varies with their sophistication. Individual clients or small businesses are generally less familiar with generative AI technology, which can result in misunderstandings and billing disputes. Conversely, sophisticated corporate clients usually have a better grasp of AI technology and its application in professional services, including legal work. These clients often have in-house legal teams that negotiate fee agreements or outside counsel guidelines that define the parameters for legal fees, including amounts and rates. Consequently, fee arrangements negotiated with sophisticated parties are less likely to be scrutinized or deemed unreasonable.⁹ To mitigate these risks, attorneys can tailor their communication and billing practices to ensure clarity and transparency, particularly for less sophisticated clients.

Conclusion

Using generative AI in legal practice offers significant benefits but raises ethical challenges, particularly in billing for work performed with AI assistance. By adhering to the principles outlined in Rule 1.5(a) and (b) and incorporating guidance from relevant ABA and state ethics opinions, attorneys can navigate these challenges ethically. Clear communication with clients and reasonable and transparent billing practices ensures that the use of generative AI in legal practice is effective and ethical.

Cassie Hanson is a legal ethics lawyer with substantial experience in the field of ethics and professional responsibility. As conflicts and ethics counsel at Fredrikson & Byron, P.A., Cassie is focused on legal ethics, conflict prevention and resolution, legal malpractice/loss prevention and trust account compliance. Additionally, Cassie worked for 20 years as a senior litigator and adviser at the Office of Lawyers Professional Responsibility. Cassie investigated and prosecuted high-level attorney discipline cases. She is an experienced trial attorney and appellate advocate who regularly argued complex attorney discipline cases in front of the Minnesota Supreme Court. She has advised various government agencies, law firms and other stakeholders on legal ethics and professional responsibility. Cassie is a frequent public speaker on ethics and lawyer well-being.

Notes

1. Florida Opinion 24-1 (January 19, 2024), p. 6.
2. D.C. Legal Ethics Opinion 388 (April 11, 2024), section E.
3. State Bar of California Standing Committee on Professional Responsibility and Conduct, “Practical Guidance for the Use of Artificial Intelligence in the Practice of Law” (November 16, 2023), p. 4.
4. Florida Opinion 24-1 (January 19, 2024), p. 6.
5. New York State Bar Association, “Report and Recommendations of the New York State Bar Association Task Force on Artificial Intelligence” (April 6, 2024), p. 58.
6. D.C. Legal Ethics Opinion 388 (April 11, 2024), section E.
7. Florida Opinion 24-1(January 19, 2024), p. 6.
8. Id.
9. For an in-depth discussion of reasonableness of fee agreements and fee agreement construction between a law firm and sophisticated corporate client, see Edward J. Cleary, Can Tobacco Suit Fees Ever Be So Large as to Be Unethical?, Minn. Law., Jul. 24, 1998.