Ethical Copying in the Artificial Intelligence Authorship Era: Promoting Client Interests and Enhancing Access to Justice

I. Introduction

Practicing lawyers are prolific writers. Their writing spans a variety of genres: they prepare internal, objective memoranda along with persuasive briefs; they draft contracts, complaints, and correspondence; and they finalize wills and expert-witness statements alike. To complete all this work, lawyers have come to rely on technology—first typewriters, then word processors. Now they are turning to artificial intelligence (“AI”) to automate some of this time-consuming work.

Although best practices call for brevity, lawyers’ written works are often voluminous. And while best practices also call for individualized tailoring, some lawyers’ “writing” involves a lot of copying, drawing heavily on the lawyers’ own past work product or words previously composed by another.^[1] Indeed, providing affordable access to justice can depend on the cost-savings of copying.

Are copying lawyers “plagiarizing”? Yes and no. Sometimes. It depends. Who’s asking?

“Plagiarism” is broadly defined, and in many ways the concept is a poor fit for much of the practical writing that lawyers create. Plagiarism is academic malpractice, and it is literary theft.^[2] But when practicing lawyers prepare contracts, memoranda, interrogatories, briefs, and other documents on behalf of clients, they are producing necessarily derivative prose that is neither scholarship nor literature. So, as a creative lawyer might argue, even a lawyer who copies without citation is not a “plagiarizer.”

Yet practicing lawyers have been criticized, and even disciplined, for “plagiarism.” How can they commit academic malpractice outside of academia? How can adopting someone else’s legal argument equate to stealing someone’s poem or novel? And if the mimicry is literary theft, is it punishable as a wrong, or worthy of praise in the name of providing affordable access to justice?

Technological advances have made copying much easier for lawyers than it once was. First, typewriters allowed both lawyers and nonlawyers to prepare and file legal documents with uniformity. Then came digital word processing and cut-and-paste tools; those tools—paired with online databases filled with primary and secondary authorities, forms, and lawyers’ briefs—allow lawyers to quickly locate effective, pre-drafted prose to incorporate into their own document.

Now AI and machine learning have arrived, further automating the generation of documents on behalf of clients. That automation is (mostly) good news. Producing legal writing is time consuming and expensive. Many clients cannot afford to pay the fees associated with bespoke legal work. And they are better served with a computer-drafted document lightly customized by a lawyer than they would be representing themselves. Still, the introduction of brief-writing software complicates the analysis of the risks and rewards of copying by lawyers.

As bots become more and more adept at drafting legal documents, the profession ought to revisit and update some of its longstanding norms. Even absent the rise of writing robots, the profession ought to be considering the incongruence of applying plagiarism ideals to practical legal writing.^[3] But the rise of AI as a drafting tool invites the profession to further rethink the balance the rules strike with respect to lawyers who copy without quoting or citing if that choice stands to enhance access to justice. This essay builds upon earlier suggestions, offering a few new ideas centered on plagiarism and the growth of automated drafting involving AI.^[4]

This essay first echoes what others have observed: the “plagiarism” label is a bad fit for any copying that practicing lawyers engage in.^[5] Next, the essay suggests that copying—even unauthorized and unattributed copying—can be helpful: it can expand access to justice by reducing the cost of preparing legal documents.^[6] It then explores how the introduction of AI in the form of brief-drafting software shifts the conversation about copying by lawyers and the application of the plagiarism label.^[7] Finally, it considers possible updates that would reflect changing practices around the efficient production of legal writing, whether through advisory opinion or changes to the Model Rules of Professional Conduct’s comments.^[8]

At its core, this essay—like the conference presentation on which it is based—invites the profession to relax the academic plagiarism norms it has long mapped onto practical legal writing. It does so by replacing inapplicable academic priorities with an access-to-justice lens. If the legal profession might chip away at the justice gap by embracing AI while prioritizing client needs over academic and literary concerns, it ought to try.

II. Background: Plagiarism & Lawyers

A. Plagiarism Defined

The definition of plagiarism is broad: to plagiarize is to use ideas or words without attribution.^[9] Anti-plagiarism rules compel students and researchers to create new works instead of copying from others—and to quote and cite carefully when they do copy—because those who lift passages without proper attribution can suffer academic misconduct charges; outside of academia, others who publish must take similar precautions to avoid reputational damage.^[10]

The concept of plagiarism aims to advance fairness in academic circles: every scholar, whether amateur or professional, must be evaluated on the basis of her own efforts. She is likely to build on the work of others, but she may do that only to the extent that her careful citations assign appropriate credit to her sources. By incorporating pre-existing language, word-for-word, without quotation marks and citation, she steals. Though it can be harder to detect, merely incorporating a pre-existing idea without a proper citation is also called plagiarism. Indeed, if the writer herself has previously published on a topic, she must cite her own earlier work to avoid self-plagiarizing.

Plagiarism also serves a purpose in the literary world, supporting copyright law’s endeavor to protect the interests of those who invest time, energy, and money in generating or publishing written intellectual property. A novelist creates intellectual property when she puts pen to paper. To replicate or build upon that prior work, one must secure the copyright owner’s permission, research and abide by the boundaries of the fair use doctrine, or wait for the copyright protection to expire.

B. Copying in the Practice of Law: Useful & Common, Yet (Occasionally) Disciplined

Practicing lawyers produce a great deal of text, and a lot of legal writing necessarily builds upon pre-existing works. That means that lawyers copy a lot—and for several good reasons.

First, lawyers copy because doing so is consistent with core principles of the legal system. A lawyer’s job is often to explain something that already has been explained. The lawyer might need to summarize a complex statute or regulation that is frequently at issue and therefore the subject of many past writings. She might need to succinctly explain a judicial decision that many other attorneys have relied upon and therefore described in writing before. Applying the principle of stare decisis, courts are to let stand that which already has been decided. But they may not know what that is until a lawyer educates or reminds them. Thus, to capably represent their clients, lawyers must present written material about the same topics again and again. Their job is not to find something new to say; it is to clearly restate what is already known.

So lawyers use the ideas of others all the time. To advance their clients’ goals, they have to copy old ideas; they root their written arguments in the already-accepted arguments of others, ideas that began as contentions and then became law.^[11]

Second, lawyers often copy the words of others because lawyers, along with judges, value precision. They strive to compose the most exact, complete, and concise language to define a legal term or explain a legal rule. If an effective passage has been drafted already, and a lawyer can identify no way in which to further tighten that language, copying it may maximize precision.^[12]

Third, and relatedly, lawyers copy because they are risk averse. There’s a place in the law for advancing creative, novel arguments. But when tested language exists, there’s little reason to try to brighten a rule explanation with novel words or punctuation. Indeed, it may constitute legal malpractice to ignore tried, true language in favor of testing a lawyer’s creative expression on the same topic.

Finally, clients appreciate value. Hiring a lawyer is different from hiring a biographer to document a life or a screenwriter to bring a novel to the stage. Clients are not investing in a writer’s speculative effort to adapt a play for the silver screen. They do not want to pay high hourly fees for a lawyer to indulge in a time-consuming creative endeavor. Rather, clients want a document drafted well, and quickly. Copying is efficient, and clients value efficiency and results over novelty or artistry.

Because lawyers, having spent seven or more years in the world of higher education, frequently copy—including without citation—plagiarism concerns have leapt from academic and literary circles to the sphere of legal writing. Judges are among those who have criticized lawyers for plagiarizing in their practical legal writing.^[13] As Professor Douglas E. Abrams explained, according to courts, “plagiarism in court filings [is] ‘reprehensible,’ ‘intolerable,’ ‘completely unacceptable,’ and ‘unprofessional.’”^[14] Lawyer plagiarism has also prompted professional disciplinary inquiries. Practicing lawyers have been disciplined for plagiarizing segments of briefs.^[15] “[P]lagiarism amounts to a misrepresentation to the court” because it involves, in the language of Model Rule of Professional Conduct Rule 8.4(c), “deceit or misrepresentation.”^[16]

Indeed, some have suggested that the prohibition against dishonesty is not the only rule of professional responsibility implicated by a lawyer’s unattributed copying; rather, Professor Abrams has argued that it is also “prejudicial to the administration of justice” and thus should subject lawyers to discipline under a separate section of the same rule, Model Rule 8.4(d).^[17] “[L]awyers’ plagiarism [in written submissions to the court] is prejudicial to the administration of justice because it creates a risk that the court’s written opinion will inadvertently plagiarize.”^[18]

One line of attack against “plagiarizing” lawyers is worthy of special attention. Practicing lawyer and former public defender J. Vincent Aprile II explained that prosecutors (who typically enjoy greater resources than public defenders) and courts have criticized criminal defense attorneys for relying on motions based on sample libraries.^[19] In other words, these lawyers and judges have weaponized ethics complaints and professional discipline—threatened under-resourced, over-extended public defenders who are working to deliver constitutionally promised representation—based on the act of copying ideas and words without proper citation.^[20] The uneven field on which these public defenders litigate becomes even more challenging if they cannot comfortably rely on stock language to quickly prepare documents that must be filed to preserve their clients’ rights.

C. Lawyer “Plagiarism”: Not the Wrong It Seems

Though plagiarism concerns have seeped into the realm of law practice, the concept of academic plagiarism is a bad fit for practical writing by practicing lawyers. As explained above, writing lawyers do—and often should—copy. Thus, scholars—people who are logically and usefully subjected to plagiarism norms—have argued that practicing lawyers should be exempted from anti-plagiarism expectations. In the eyes of these scholars, “plagiarism” is not only justifiable in legal practice; it is necessary.

For instance, Professor Andrew M. Carter argues that a rule barring plagiarism in the practice of law rule “cannot be justified,” explaining that such a rule is especially problematic because it can be cost-inefficient for the client.^[21]

Professors Megan E. Boyd and Bryan L. Frye go farther: They begin their article boldly, turning on its head the notion that plagiarism in the practice of law is unethical.^[22] In their view, reluctance to plagiarize—not plagiarism—is the problem: “[a]s a practicing lawyer, if you aren’t plagiarizing, you’re committing malpractice.”^[23]

These arguments are compelling. The focus of a legal writer should be efficiently and effectively explaining the applicable law and the arguments it supports. Her highest priority should be articulating legal principles and arguments with clarity and precision, helping to resolve disputes while also loyally guarding her client’s interests. That includes minimizing the client’s expense. Scholars—wordsmiths who are not billing clients for their work—can choose to experiment with language endlessly, or at least until their publisher’s deadline. But lawyers need to focus on cost-effective advocacy aimed at protecting their clients’ rights and advancing their interests. Tools for cost-effective advocacy are especially critical in the context of today’s access-to-justice gap.

III. Access to Justice: A Crisis that Could Be an Opportunity

The access to justice crisis is well documented.^[24] It has been discussed widely in legal scholarship, by the popular media, and at legal-aid fundraisers. The crisis is so familiar that it should no longer be surprising. And yet, reminders about the relative inaccessibility to legal representation suffered by the people with the most acute needs tends to drop jaws.

For instance, one study concluded that civil legal-aid programs in the state of Massachusetts turn away almost two-thirds of eligible clients facing serious issues, like eviction, denial of benefits, or lack of child support, due to lack of funding and staffing; an overwhelming 80% of cases involving family law were turned away.^[25] In Oregon, the Department of Justice reports that just 29% of spouses in divorce cases are represented by lawyers; a meager 17% of residents and landlords benefit from legal representation in eviction cases.^[26] Although some choose self-representation, many lack the financial resources to hire an attorney. Finally, low-income Californians sought and received legal help for about thirty percent of the problems they experienced, even though all those problems were potentially legally actionable.^[27]

Those who continue to spotlight the struggle of non-affluent Americans to secure professional help in navigating legal problems seek to remedy this inequity through a variety of strategies. One strategy is to decrease the demand for legal help: simplifying laws and legal processes can minimize, or even eliminate, the need for professional legal assistance in some situations. Another strategy is to increase supply: creating paraprofessionals who are prepared and licensed to assist with some categories of legal disputes can lower the cost of some types of professional legal assistance.^[28]

One more way to lower the cost of professional legal assistance is through the efficiency offered by technological advances like cutting and pasting from digital brief banks—but only if the profession agrees that copying is acceptable. If lawyers can copy the work of others without worrying about a disciplinary complaint based on plagiarism, they can deliver much more cost-effective legal services to their clients.

In other words, applying an access-to-justice lens to an evaluation of the copying engaged in by lawyers might help justify, and thus expedite, the erosion of traditional anti-plagiarism rules.

IV. Artificial Intelligence Changes . . . Everything?

Enter the robot brief writers. Do they change anything? Might they change everything? Yes. No. It depends. Who’s asking?

When a legal writing professor learns about Compose, a program that allows legal writers to create a first draft of a brief using drop-down menus, she might be caught off-guard. Her first reaction might be disbelief. The way many legal writing professors teach and engage in the craft of legal writing, the endeavor requires not only technical knowledge but also a feel for language—an ear for style, tone, and flow that helps a writer achieve both clarity and persuasion. How could a machine craft something as nuanced as a legal argument? The professor might think about the inaccuracy of the word suggestions offered by her AI-powered email and texting platforms. She might likewise reflect upon the awkward articles she has encountered online that were written by bots.

But listening to the first-ever robot-written stand-up comedy routine might change her mind.^[29] AI is next-level powerful. The computer’s grammar and word choice make the writing feel rough; it could use the smoothing hand of a human editor. And there’s a bit of human je ne sais quoi missing from the work.^[30] But still. A robot wrote a stand-up comedy routine. Not only is it understandable; it’s also funny. This technology is coming along quickly. Its sophistication and capacity are not to be underestimated.

Yet in considering this technological advancement, the professor might sense hypocrisy, conflict, or seeds of inequity. The professor is teaching her students to craft without the head start of a brief bank or computer program, despite knowing that those students will likely use those tools once they enter practice—if resources permit. Moreover, the bots are, in effect, replacing the junior associates that the students aspire to become, working at high-priced firms that serve mainly corporate or wealthy individual clients. Those are the clients who stand to benefit most from a Compose first draft, receiving discounted drafting services that are still so expensive that they likely remain out of reach for clients of modest means. Will the profession continue to worry about the unattributed copying of pre-existing language in briefs filed with a court? Or might this technology be the development that cures some members of the profession of their continued, and misplaced, devotion to that principle?

Preliminary thoughts on those topics appear elsewhere.^[31] To begin, because the profession is going to allow lawyers to lean on paid ghost writers—like uncredited junior associates, judicial law clerks, and now artificially intelligent machines—the profession should let go of any “plagiarism” concerns linked to the use of samples, brief banks, and other stock language. If borrowing—without attribution—language that someone (or something) other than the attorney of record drafted is acceptable in some circumstances, it ought to be acceptable in all circumstances.

The 2021 LWI Symposium on Artificial Intelligence & the Legal Profession presented an opportunity to reiterate and expand upon this idea: the growth of AI invites the profession to revisit concerns about copying by lawyers. Computers are not just word processors, absorbing lawyers’ keystrokes and flagging errors in grammar and spelling; they are now drafters: selecting, organizing, and expressing content. This technological advancement may not change everything, but nevertheless it invites a reexamination of some things most lawyers take for granted, including the rules that apply to copying.

A. Bot Boilerplate Is Nothing New

The idea of plug-and-play transactional documents seems relatively uncontroversial. Boilerplate by bots is no big deal. In part, that is because automation in the creation of documents like this is old news. Long before lawyers used AI to efficiently create legal documents like contracts, lawyers relied on forms for the same purpose. The same is true with respect to pleadings and other formula-driven or template-based written works. That robots can automate the work formerly completed by junior attorneys and formbooks combined is relatively easy to accept. It layers thoughtful customization—the careful thinking and revising lawyers are trained to complete—atop a copied base.

That said, and purely anecdotally, some lawyers have been known to negatively judge their peers who circulate transactional documents that are obviously based on publicly available templates like bar practice guides.^[32]

B. But Bot Briefs Seem Different

At least one platform, Compose, now allows lawyers to compose the first draft of a brief via drop-down menus.^[33] Based on jurisdiction, claim, and issues, a lawyer can populate a document with an explanation of the applicable law: rules and case illustrations; quotes and citations. It’s like an extremely extensive, automatically updated brief bank. So now technology allows lawyers to create first drafts of briefs as quickly as they can generate a first draft of a transactional document; moving from first to second draft will involve analytical and strategic thought, but Compose will short-cut much of the time-consuming work of selecting, synthesizing, excerpting, and organizing authorities that is traditionally involved in the writing process.

The profession now has not only Bot Boilerplate but also Robotic Rhetoric.

Log onto Compose’s website to see how it pitches itself to potential customers. As of now, the website’s large text lures users with this promise: “Craft exceptional briefs, without the busy work” by using the platform’s “all in one research and drafting tool.” That “tool provides all the arguments, legal standards, and pre-packaged research you need to get things done, faster than ever.”^[34]

It’s not all technology doing the behind-the-scenes work; lawyers are involved as well. The service “offers intelligent drafting . . . [b]y combining the skills of experienced litigators with groundbreaking machine-learning technology.” It’s a time-saving tool that helps writers get things right and get them written. And if the motion the user needs is not yet in the program’s brief library, the user can submit a request to have it added.

This development is likely to be embraced and celebrated by some while leaving others feeling unsettled. Some of that discomfort likely reflects doubt about the ability of AI to write well. And some of that discomfort relates to what lawyers were taught as students about cheating, transparency, and fair play. Plagiarism is a no-no for students; if a law student paid Compose $1,500 for a first draft of an appellate brief for credit, she could be sanctioned for academic dishonesty. Plagiarism. But only some plagiarism is off-limits for practicing lawyers, who—unlike the students they once were—are permitted to lean on others while preparing their written work.

Tom Goldstein of SCOTUSblog said that Compose “promises . . . to usher in a new era in legal briefing.”^[35] Perhaps it will also be the innovation that prompts the profession to rethink its long-standing objections to copying. Lawyers will not be expected to credit Compose as a source in order to avoid an ethical complaint for plagiarizing a brief, just as they are not expected to credit a paid contract attorney to whom they delegated drafting work. There may be no better argument regarding how poorly plagiarism norms fit the lawyering profession than the introduction and marketing of Compose: the next great thing in legal writing for a law practitioner—a tool that efficiently delivers effective written advocacy with little user effort—is a tool that might get a law student suspended or expelled.

And while Compose may transform the legal writing process for some segments of the legal profession, its current pricing model means that many lawyers, like those in under-funded legal aid and public defender offices, will keep relying on their own brief banks instead.

To the extent that AI can chip away at the access-to-justice crisis by making legal help more available and affordable, it’s something to embrace, celebrate, and facilitate. But to effectively achieve that goal, the tool needs to be affordable for clients of limited means. Short of that, its use should at least silence any continued criticism of brief-bank reliance.

V. New Considerations for a New Era

The Model Rules of Professional Conduct seek to balance competing interests. As the basis for state legal ethics codes, the rules offer lawyers guidance when they feel pulled in multiple directions. For example, the rules help resolve the tension that exists when a lawyer is driven to act swiftly by one client’s interest but to wait and see by another’s, or when the lawyer is expected to preserve client confidentiality but must also honor a duty of candor to the court. Similarly, a conflict arises when a lawyer’s desire to earn high wages collides with the profession’s reputational concerns and clients’ strong interest in receiving high-value representation—the best possible lawyering at the lowest possible price. The principles reflected in the rules often don’t tell a lawyer exactly what to do; rather, they relate a standard and ask lawyers to use their judgment in applying it.

Several provisions of the rules of professional conduct are arguably affected by AI writing tools. These include the rules addressing misrepresentation, competence, diligence, fees, independent judgment, and supervisory duties, each of which is addressed below.^[36] State bar organizations, along with the ABA, could address this by issuing a formal opinion on the ethical implications of artificial intelligence. Further, though more complicated, the ABA and state courts could re-examine each rule and revise its associated commentary in light of the introduction of brief-writing software and other available AI drafting assistance.

A. Misrepresentation

The most important rule to focus on is probably the one that lawyers have been disciplined for violating by “plagiarizing.” As noted above and discussed at length elsewhere, lawyers who have copied without proper attribution have been criticized for violating the rule prohibiting misrepresentation—specifically, for presenting another’s ideas or written work product as their own.^[37]

The same critique might be applied to someone who files a document automatically drafted by brief-writing software. If incorporating language from a brief bank can be criticized as an unethical misrepresentation, drafting via a computer’s drop-down menu could as well.

A new comment to Model Rule of Professional Conduct 8.4 (or other rules, as appropriate) could address this new issue:

Reliance on or incorporation of pre-drafted or co-drafted materials, from or with any accurate source, to deliver competent and affordable legal representation does not alone amount to misrepresentation in violation of the rules, even if the pre-existing writing, human co-author, or AI resource is not cited. The question is whether a competent lawyer acting in good faith would regard the copying as facilitating the delivery of cost-effective legal representation.^[38]

This comment prioritizes clarity, accuracy, and efficiency in the explanation of legal principles over revelation of the original composer; while that information can be instructive, it often is not important at all.^[39]

B. Competence & Diligence

Another core duty implicated by the proliferation of AI tools is competence. In 2012, the ABA updated the comments to Model Rule of Professional Conduct 1.1, Competence, to explicitly mention technology expertise: “To maintain the requisite knowledge and skill, a lawyer shall keep abreast of changes in the law and its practice, including the benefits and risks associated with relevant technology . . . .”^[40] Most states have adopted this language or language like it.^[41] It requires that lawyers stay up-to-speed with important developments like those involving e-discovery and e-filing, data storage and privacy, online legal research, and e-mail and remote-meeting tools.

The growth of computer-written legal arguments invites consideration of additional revision to the Model Rules’ comments, related perhaps not only to the Rule 1.1 competence rule but also to the Rule 1.3 diligence rule. Language along these lines could be added to the comments associated with either of those rules:

[Competent/diligent] lawyers may reasonably rely on stock language located in a form or generated by AI where the provided language accurately reflects the law and adequately and efficiently addresses the client’s needs.

Comments like that would emphasize that bar authorities should discipline lawyers for submitting a computer’s draft only if doing so reflects a lack of competence or diligence; that might occur where a lawyer failed to proofread or customize computer-generated stock language or where the lawyer’s procrastination prompted inappropriate reliance on a raw computer draft in need of revision. In short, the rules could make plain that reliance on trustworthy AI, including in drafting, is reasonable and consistent with both competence and diligence.

C. Fees

The rules of professional responsibility protect clients by prohibiting lawyers from charging or collecting unreasonably high fees or expenses; and what is reasonable is measured, in part, by the customary charge for similar services.^[42] The customary charge will change as the use of AI increases; according to Compose, its motion library provides “[a]ll the arguments, legal standards, and precedent you need to draft stronger motions 4x more efficiently.”^[43] If true, the platform stands to materially shift norms around the typical process and resulting expenses associated with preparing a brief. That change creates risk for lawyers, whose fees might be challenged as unreasonable under the rules of professional conduct.

For instance, these and other objections may be raised by clients who are reviewing an itemized bill: (1) with respect to costs,^[44] $1,500 for a computer-generated first draft is unreasonable because the amount is too high (compared, for instance, to pulling a first draft out of a firm’s in-house brief bank for free), and (2) with respect to fees, any amount of attorney time billed for basic legal research and a human-created first draft is unreasonable when AI automatically generates solid work product and protects against some types of human error as well.

To address those concerns, the ABA could add to Rule 1.5, which prohibits unreasonable fees, a comment along these lines:

Lawyers may charge reasonable fees and costs for locating, analyzing, and incorporating stock language or for generating and customizing an AI-created draft. Reliance on reasonably priced AI can be appropriate, though AI will not be an available or appropriate substitute for human drafting in some instances.

In crafting a new comment to Rule 1.5, along with other comments, finding an appropriate balance between generality and specificity—so that the comment offers both useful guidance and flexible application to a broad range of factual scenarios—will be challenging. Indeed, that challenge suggests that a formal opinion may be a more appropriate and expedient strategy for offering guidance to lawyers regarding the many implications of using AI as a drafting tool.

D. Independent Judgment, Communication, & Client Deference

Determining when and how to rely on computer assistance to draft legal documents will require lawyers to communicate with their client, consistent with Model Rule 1.4, and then “exercise independent professional judgment,” as required by Model Rule 2.1. These provisions present another opportunity for the ABA to add a clarifying comment to Model Rule 1.2, Scope of Representation and Allocation of Authority, explicitly mentioning that the use of AI may be impliedly authorized to carry out the representation:

A lawyer must use her professional judgment, after consulting with her client as required by Rule 1.4(a)(2), to determine the extent to which reliance on AI drafting tools is consistent with the client’s needs and goals.

To the extent the attorney-client consultation reveals that lawyer and client have different opinions about using AI tools, Rule 1.2 suggests that the lawyer’s professional judgment should probably prevail; that is because whether to use AI for drafting purposes is not an objective but rather a means or strategy by which the client’s objectives might be pursued.^[45] But the existing comments create some ambiguity: while clients normally defer to lawyers on strategic and technical matters, lawyers ordinarily defer to clients on questions of expense.^[46] Thus, a revised comment could provide useful guidance.

E. Supervision

Under Model Rules 5.1 and 5.3, lawyers are responsible not only for their own conduct, but also for the conduct of the lawyers and nonlawyers they supervise. These supervisory duties raise some interesting questions for lawyers who turn to technology to provide not only the answers to their legal research inquiries, as they long have with platforms like Westlaw and Lexis, but also for written work product. Compose claims to sell lawyers not just algorithm access to help them search for and sort materials with editorial enhancements, but rather “[e]very argument [and] legal standard . . . written and researched by experienced attorneys.”^[47] So there’s a lawyer relying on a bot relying on the vendor’s lawyers. How much transparency exists to make oversight possible? How much deference or trust is appropriate?

Rule 5.3 states that when a lawyer retains or associates with a nonlawyer, the lawyer must “make reasonable efforts to ensure that the person’s conduct is compatible with the professional obligation of the lawyer.”^[48] The lawyer can be disciplined if she orders, ratifies, or fails to remedy conduct by the nonlawyer that would be professional misconduct for a lawyer.^[49] The comments accompanying this rule reference “secretaries, investigators, law student interns, and paraprofessionals” and specify that these “assistants, whether employees or independent contractors, act for the lawyer” in providing legal services.^[50] The comments likewise mention nonlawyers who work outside the lawyer’s organization, like investigators, document-storage-and-management companies, and printers and scanners.^[51] The lawyer is, on some level, responsible for ensuring that all these partners act consistently with the lawyer’s professional obligations.

But it seems unlikely that the lawyer can benefit from the efficiency offered by a bot while also personally confirming that the bot has not engaged in any unauthorized copying. Here, as in the other areas discussed above, the rule drafters might add commentary acknowledging that lawyers’ reliance on ghost-writing lawyer/programmers—and the technology they have developed—creates a new category of supervisees parallel to the document management companies and third-party printers already explicitly mentioned. For instance, AI legal service providers could be added to the list of people and entities outside a lawyer’s firm that the lawyer must reasonably oversee to avoid accountability for bot behavior, with special emphasis on what is reasonable under the circumstances.

VI. Conclusion

Scholars in many fields are focused on the myriad effects of the rise of AI. As the 2021 LWI Symposium on Artificial Intelligence & the Legal Profession has made clear, the legal discipline specifically is shifting rapidly, and in a variety of ways, as new technology emerges and takes hold. Legal academics exploring the impact of AI are excited about its potential for expanding access to justice.^[52]

The intersection of legal writing and professional responsibility requires some analysis if AI drafting tools are to help address the justice gap. This essay begins to assess how the legal profession might adjust various professional responsibility standards so that the profession is better poised to capitalize on the efficiencies promised by emerging technology. These ideas are intended to advance a conversation about how the discipline might, through rule interpretation in a formal opinion or revised commentary, better prioritize client interests and facilitate access to justice.

Brief-writing technology will continue to evolve, and its use will grow. As that growth occurs, “copying” by machines and the humans who rely on them will increase and, so long as we permit it, access to justice should rise. The continued tension between anti-plagiarism expectations and the realities of modern law practice deserves attention and resolution, as policy has been out-of-step with practice and client needs for some time.

This new era calls for deep reflection. If the profession can relax old norms to better leverage the power of technology to efficiently create clear, precise writing, it should. But the profession should also work to ensure that this technology serves to close, not widen, the justice gap. That could mean announcing a pro bono donation expectation for brief-writing software companies. Or it could mean introducing a “buy one, give one” norm to law firms’ use of brief-writing AI. Or maybe it means encouraging the introduction of AI brief-writing competition to help drive down the cost of access to AI-crafted briefs.

The introduction of AI legal-brief drafting presents a moment of opportunity for the legal profession. Because bots can draft briefs, the profession must revisit longstanding assumptions about authorship and critiques of copying. It must rethink what its rules suggest and how they apply. And it should center all this reflection and reinvention on the lofty aspiration of expanding access to justice.