Ethics in Brief – Bias Beware: How the Use of AI Tools in Legal Practice May Create Unintended Bias and Potentially Violate California Rule of Professional Conduct 8.4.1

By: Alara Chilton

AI tools have grown rapidly over the last few years within the legal profession and the business sector as a whole. The number of AI^[i] tools will likely continue to grow as more law firms and lawyers utilize AI tools with the aim of creating greater efficiency and profitability.

When utilizing any AI tool, however, a lawyer must be mindful of the Rules of Professional Conduct and the State Bar Act. While the use of AI can implicate various ethical duties^[ii], this article focuses specifically on the issue of bias as it relates to the use of AI and how unintended bias may potentially implicate California Rule of Professional Conduct 8.4.1.Consider the conversation below between two personal injury lawyer colleagues:

Vanessa: I have really been impressed by how AI is making my practice more profitable.

Sarah: How exactly are you using AI to make your practice more profitable? I would be interested in learning how I could do the same with my practice.

Vanessa: One area where I find AI particularly useful is in analyzing my firm’s historical client data. Recently, I entered client data from the past five years, including clients’ ZIP codes, litigation costs, and the amounts ultimately recovered. The AI analysis revealed two important facts which surprised me: (1) clients in ZIP code A produced at least fifty percent more profit than clients in other ZIP codes; and (2) clients in ZIP Code B produced only approximately five percent of my firm’s profits. 

Sarah: That is very interesting information. I am curious—do you know anything about the demographics of ZIP codes A and B? 

Vanessa: Yes—my AI tool looked into the demographics of both ZIP codes. Its research revealed that ZIP code A was approximately eighty percent White, ten percent Latino, and ten percent Asian or Black. ZIP code B was approximately eighty percent Latino, ten percent White, and ten percent Asian or Black. Since AI is known for making mistakes, I verified the demographics using census information. 

Sarah: So how did you use this historical firm data to increase your profit margins?

Vanessa:  I started by having my AI tool use this information when screening prospective clients so that my firm gives top priority to signing prospective clients from ZIP code A.

Sarah: How exactly does your AI give top priority to prospective clients from ZIP code A?

Vanessa: I don’t automatically reject cases from other ZIP codes. Instead, I had my AI tool create a Prospective Client Scoring (“PCS”) system based on various criteria, such as the estimated case value, absence of prior similar injuries, and liability strength. And every prospective client who lives in ZIP code A receives additional points on their PCS system score.

Sarah: Have you considered how using the last criterion to screen prospective clients could create bias that may violate Rule of Professional Conduct 8.4.1 [Prohibited Discrimination, Harassment and Retaliation]?

Vanessa: That sounds a bit far-fetched since I did not intend to discriminate against anyone—let alone on the basis of race. Isn’t that obvious since I never provided my AI tool with racial client data or made race a criterion in my firm’s PCS system?

Sarah: I know you did not intend to discriminate, but your PCS system has provided additional points to every prospective client who lives in ZIP code A. By doing so, it has given greater priority to prospective clients who live in a predominantly White ZIP code over prospective clients who live in a predominantly Latino ZIP code. This has created racial bias in your PCS system.

Vanessa: I’m surprised by this outcome because racial bias was not my intention. But does the existence of racial bias mean I am unlawfully discriminating under Rule 8.4.1^[iii]?

Sarah: Rule 8.4.1(a)(1) applies when a lawyer is representing a client or deciding whether to refuse representation. It provides that a lawyer shall not “unlawfully discriminate against persons on the basis of any protected characteristic^[iv],” including race.

Here, your firm has been using the PCS system to help you decide whether to refuse representation, and as we just discussed, the system contains racial bias in that decision-making process. The racial bias may violate Rule 8.4.1(a)(1) if you have unlawfully discriminated against others on the basis of race. However, pursuant to subdivision (c)(3) of this rule, “unlawfully” shall be determined by applicable state and federal statutes and decisions. So the fact that your PCS system contains racial bias does not necessarily mean that you have engaged in unlawful discrimination. You would have to refer to state and federal statutes, as well as case law to answer that question.

Vanessa: This conversation has made me aware of how easy it is to use an AI tool in a way that creates bias when representing or refusing to represent a client. And I am still surprised that this can happen even when a lawyer never intended to create bias or discriminate against anyone.

Sarah: Indeed. Even when a lawyer does not instruct an AI tool to consider race or another protected characteristic, the use of an AI tool can still create unintended bias and potentially result in unlawful discrimination under Rule 8.4.1. As the use of AI tools continues to grow in legal practice, lawyers must understand how unintended bias can arise and potentially implicate Rule 8.4.1 in order to protect not just their clients, but also their bar cards.

^[i] “Artificial intelligence” is defined under California Government Code section 11549.64 as “an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments.” Additionally, the proposed amendments to California Rule of Professional Conduct 1.1, Comment [1], if adopted by the California Supreme Court, will incorporate this statutory definition of artificial intelligence by reference.

^[ii] For a discussion of the various ethical duties that arise when using AI, see the updated revisions to the State Bar of California, Board of Trustees, Practical Guidance For the Use of Generative Artificial Intelligence in the Practice of Law (May 2026).

^[iii] State Bar of California, Board of Trustees, Practical Guidance For the Use of Generative Artificial Intelligence in the Practice of Law (May 2026) discusses AI bias in the context of Rule 8.4.1. Rule 8.4.1, however, does not itself prohibit “bias.” Rather, the rule prohibits unlawful discrimination, harassment, and retaliation.

^[iv] This article focuses on the application of subdivision (a)(1) when a lawyer is representing a client or deciding whether to refuse representation. Subdivision (b)(1)(i) of rule 8.4.1 also prohibits a lawyer, “in relation to a law firm’s operations,” from “on the basis of any protected characteristic,” “unlawfully discriminat[ing] or knowingly permit[ting] unlawful discrimination.” Subdivision (b)(1)(ii) of the rule further prohibits a lawyer, “in relation to a law firm’s operations,” from “on the basis of any protected characteristic,” “unlawfully harass[ing] or knowingly permit[ting] the unlawful harassment of an employee . . . .” Subdivision (c)(1) defines “protected characteristic” to include race and several other characteristics.