Nota bene: This article addresses ABA Model Rule 8.4(g), the ABA’s somewhat controversial anti-discrimination rule, which unlike many Model Rules has NOT been widely adopted by state bars and other jurisdictions, including DC. DC retains Comment [3] to D.C. Rule 8.4, providing that discrimination on the basis of race, sex, religion, national origin, disability, age, sexual orientation, or socioeconomic status may constitute serious interference with the administration of justice. Other states, like Maryland, have adopted or kept a pared down version as a rule. See MD Atty’s Rule of Professional Conduct 8.4(e) (stating a rule similar to DC’s Comment [3]). Check your jurisdiction’s specific rules and seek legal advice as needed.
The ABA’s most recent legal ethics opinion, Formal Opinion 517, attempts to clarify a criminal defense attorney’s ethical duties in the jury selection process in view of ABA Model Rule 8.4(g), which precludes discrimination (at least in jurisdictions where it has been adopted) on the basis of 11 different categories, as stated in full at the end of this article.*
For legal ethics nerds (or at least for me), the opinion is notable for being what may be the first attempt at applying an ethics rule to a specific use of AI. While various jurisdictions have addressed lawyers’ uses of AI more generally (e.g., DC Ethics Op. 388 and ABA Formal Op. 512), ABA Opinion 517 addresses the specific use of AI in jury selection under Rule 8.4, along with other non-AI issues. And in doing so, at least for me, it raises more questions than answers and potentially worsens the ethical quandary for a criminal defense attorney dedicated to the zealous defense of their client.
Peremptory Strike Law and Model Rule 8.4(g)
Criminal defense attorneys are well familiar with the basic rule from Batson v. Kentucky, 476 U.S. 78 (1986), and its progeny: an attorney may not peremptorily strike prospective jurors based on certain criteria, like race or gender. Not all discrimination is precluded, however—an attorney may strike jurors based on things like age or disability, at least in federal court, because such characteristics are not within the protections of the Fourteenth Amendment.
And nowadays, the jury selection process—especially in high stakes trials—is supported by an array of strategic tools, including jury consultants, mock trials, focus groups, and now, software tools with artificial intelligence capabilities.
Meanwhile, ABA Model Rule 8.4(g)* purports to preclude discrimination by lawyers on the basis of a wide array of 11 different criteria, many of which—like age and disability—are permissible grounds for peremptory challenges. Notably, Model Rule 8.4(g) includes an “escape valve,” permitting discrimination on the basis of the 11 categories if it is part of “legitimate advice or advocacy,” and if it is otherwise consistent with the Model Rules. The $64 million dollar question has always been, what is “legitimate advice and advocacy.”
In Opinion 517, the ABA attempts to reconcile the broad language of Model Rule 8.4(g) with a criminal defense attorney’s obligation to zealously advocate for his or her client. It’s somewhat unsurprising conclusion? Discrimination in selecting jurors that a lawyer knows or reasonably should know is precluded by law (such as Batson) is not “legitimate advocacy” and so is not permitted under Model Rule 8.4(g). If it’s not precluded by substantive law like Batson, however, it’s fair game and not a violation of Model Rule 8.4(g).
A Lawyer’s Duty When Taking Input from Others in the Jury Selection Process
The more interesting part of the opinion, though, is what it says about a lawyer’s responsibility when receiving input from others, such as the client or a jury consultant or—that’s right—artificial intelligence. In such cases, at least in jurisdictions that have adopted Model Rule 8.4(g), the opinion states that a lawyer may not follow a client directive, a consultant’s advice, or AI guidance if he or she knows or reasonably should know that doing so would constitute unlawful discrimination.
Moreover, when a client’s or consultant’s proposal is or is likely to be discriminatory in an unlawful way, the lawyer has a duty to inquire as to whether there is a sincere, legitimate reason that supports the juror challenge. As the opinion states, “if a reasonably competent and prudent lawyer would know that the reasons are pretextual, and that the proposed peremptory challenge is unlawful, then the lawyer must refrain from relying on the client or consultant.” In other words, following the client’s direction or a consultant’s advice does not give a lawyer safe harbor; a lawyer is essentially responsible for rooting out unlawful discrimination in the jury selection process proposed by others.
The Duties of a Lawyer Using AI in the Jury Selection Process
More troubling is that the ABA says a similar duty of inquiry exists when using AI to assist with jury selection. Because, it says, it is “conceivable” that an AI program could suggest striking jurors in unlawful ways, “lawyers should conduct sufficient due diligence to acquire a general understanding of the methodology employed by the juror selection program.”
While this is a valiant attempt by the ABA to start addressing the role of AI in specific contexts (see Formal Opinion 512 for general AI guidance), this particular guidance strikes me as unhelpful at best and potentially harmful to the justice system in its vagueness.
The first issue I see is that the opinion’s use of “conceivable” is telling—indicating that the authors are not even sure whether (much less how) an AI program could or would discriminate unlawfully in the juror selection process. It would be better to await further study of such software.
Second, the call to “conduct sufficient due diligence” is vague and unhelpful. My lay understanding is that AI algorithms do not lend themselves easily to obtaining a “general understanding of the methodology” being used. These are extraordinarily complex calculations—that’s what makes them AI. The opinion provides no specifics as to how to overcome this issue. Discovering bias in a particular AI program would probably take serious scientific study, something a criminal defense lawyer under the pressure of trial does not have time for. Inserting a general due diligence requirement for something as complex as AI—as an almost throwaway line—puts a lawyer in the difficult position of either abandoning the AI tool or going forward at risk of “knowingly” engaging in unlawful juror discrimination due to not conducting “sufficient due diligence.” The former shortchanges the client in a meaningful way. While the anti-discrimination tenets of Model Rule 8.4(g) are admirable in many ways, limiting the defense in a criminal case is a serious issue that should not be taken lightly.
But this raises an even bigger question. At some point, fulfilling your duties of competence and zealous advocacy for your client will require you to use AI—either in jury selection or otherwise. Using AI—even if not required—may give your client an advantage that you should provide as their counsel. But the opinion, by assigning responsibility for an AI’s potential discrimination, may be handicapping lawyers from doing the best for their clients and sacrificing zealous advocacy to overcompensate on the discrimination issue.
* * *
In sum, while the opinion is helpful in clarifying the line between discrimination that violates the Model Rule versus what doesn’t (whether or not it’s otherwise unlawful under Batson and progeny), and in clarifying an attorney’s responsibilities when taking input from others, it is unhelpful in its attempt at setting a lawyer’s duties when using AI—something that will more and more become the norm. The opinion’s guidance does little to assist lawyers grappling with these issues, raises more questions than answers, and could create an atmosphere that hampers an attorney’s ability to do the best for their client in a lawful way.
________________________
* Model Rule 8.4(g) provides in full:
It is professional misconduct for a lawyer to engage in conduct that the lawyer knows or reasonably should know is harassment or discrimination on the basis of race, sex, religion, national origin, ethnicity, disability, age, sexual orientation, gender identity, marital status or socioeconomic status in conduct related to the practice of law. This paragraph does not limit the ability of a lawyer to accept, decline or withdraw from a representation in accordance with Rule 1.16. This paragraph does not preclude legitimate advice or advocacy consistent with these Rules.
Comment [5] to Model Rule 8.4 clarifies: “A trial judge’s finding that peremptory challenges were exercised on a discrimination basis does not alone establish a violation of paragraph (g).”
Bill Zapf is a litigator and legal ethics lawyer. His practice encompasses complex civil litigation, white-collar criminal defense, government investigations, and appeals. He has a particular focus and expertise in matters involving legal ethics, including legal malpractice and attorney disciplinary matters.
July 17, 2025
Learn more about our work in Higher Education Law
