Rule 1.1 Comment 8 and AI: The Technological Competence Duty Every Firm Now Faces

Competence has always meant more than knowing the law. Since 2012, the comment to the rule on competence has said so in plain terms, and the arrival of generative AI has turned what once read as an aspiration into a concrete operating standard for every firm that drafts, researches, or reviews with these tools.

Where the duty comes from

The duty of technological competence is not a freestanding rule. It lives in a comment to ABA Model Rule of Professional Conduct 1.1, the competence rule. In 2012, on the recommendation of its Ethics 20/20 Commission, the ABA House of Delegates amended Comment [8] to Rule 1.1. The amendment inserted a clause that changed the baseline. The comment now provides that, to maintain the requisite knowledge and skill, a lawyer should keep abreast of changes in the law and its practice, including the benefits and risks associated with relevant technology, and should engage in continuing study and education.

Two features of that language matter for AI. First, it is framed around both benefits and risks, so the duty is not satisfied by either reflexive adoption or reflexive avoidance. Second, it is tied to "relevant" technology, which is contextual. Technology a litigator must understand may differ from what a transactional lawyer must understand, and the relevant set changes over time. What was once a debate about email and e-discovery is now, unavoidably, a question about large language models.

This is the heart of the Rule 1.1 Comment 8 AI question. The comment does not name any technology, and it did not need to. It set a standard of keeping current, and generative AI is now squarely within the technology a competent lawyer is expected to understand at a working level.

How widely it has been adopted

The Model Rules are a template, not binding law. Each jurisdiction adopts its own rules of professional conduct, usually patterned on the Model Rules. So the practical question is not what the ABA says, but what your jurisdiction has enacted.

On that point the trend is decisive. According to the tracker maintained by legal-technology writer Robert Ambrogi at his LawSites site, which has followed adoptions state by state since the amendment, 40 states have adopted a duty of technological competence, along with the District of Columbia and Puerto Rico. The District of Columbia approved its version in 2025, and Puerto Rico followed, with its rule effective at the start of 2026. Most adopting jurisdictions took the ABA language close to verbatim, though some modified it.

For a firm with lawyers admitted in several states, the takeaway is straightforward. In most of the country the duty is now express in your own rules, not merely a national recommendation. Even in the handful of states that have not formally adopted Comment [8], the underlying competence rule still requires you to handle a matter with the thoroughness and preparation reasonably necessary, which is difficult to do with a tool you do not understand. The safer assumption is that the duty applies to you and that AI is within its scope.

What the ABA says competence with AI requires

The leading authority is ABA Formal Opinion 512, issued by the Standing Committee on Ethics and Professional Responsibility on July 29, 2024. It is the ABA's first comprehensive ethics guidance on generative AI, and it reads the existing rules onto the technology rather than inventing new ones. It addresses competence, confidentiality, communication with clients, candor toward the tribunal, supervision, and fees.

On competence specifically, Opinion 512 makes two points worth holding onto. First, a lawyer need not become a machine-learning engineer, but must have a reasonable understanding of the capabilities and limitations of the specific tools being used in a representation. Second, because these tools change quickly, that understanding has to be kept current rather than acquired once. The opinion is candid about the defining limitation: generative AI can produce output that is plausible, fluent, and wrong, including invented citations and quotations.

The opinion also draws lines that catch firms by surprise. On confidentiality, it cautions that lawyers must understand how a given tool handles input data before entering client information, and it states that generic, boilerplate consent buried in an engagement letter is generally not enough to authorize disclosing client confidences to an AI tool. On fees, it explains that a lawyer ordinarily may not bill a client for the time spent learning to use AI tools generally, the same way you would not bill a client to learn basic legal research. The throughline is that AI is governed by duties you already owe, applied to a new instrument.

Why hallucination is the competence problem

Understanding the limits of the technology is not abstract. The single most important thing a competent lawyer must understand about a large language model is that it generates text predicted to be responsive, not text verified to be true. It can fabricate case names, docket numbers, quotations, and holdings with complete confidence. The industry term is hallucination, and it is not an occasional glitch. It is a property of how these systems work.

The courts have already supplied vivid illustrations. In Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023), Judge P. Kevin Castel sanctioned two attorneys and their firm after they filed a brief containing fictitious cases that ChatGPT had generated, complete with fabricated quotations and citations to nonexistent decisions. In an order dated June 22, 2023, the court imposed a $5,000 sanction under Rule 11 and found that the lawyers had acted in bad faith, in part because they continued to defend the fake authorities even after their existence was questioned. The fabricated material, the court noted, included reasoning it described as gibberish.

The appellate courts have followed. In Park v. Kim, 91 F.4th 610 (2d Cir. 2024), the Second Circuit confronted a reply brief that cited a case the attorney admitted she had produced using ChatGPT and that did not exist. The court referred the attorney to its Grievance Panel, explaining that the obligations imposed by Rule 11 require attorneys to read and confirm the existence and validity of the authorities on which they rely. The lesson is that AI assistance does not dilute the duty to verify. It concentrates it.

These are not isolated incidents anymore. Courts across the country have issued similar orders, and some have entered standing orders requiring disclosure or certification regarding AI use. A competent lawyer treats every AI-generated citation, quotation, and factual assertion as unverified until checked against the primary source.

What this means for a small or midsize firm

The duty does not require your firm to avoid these tools. Used carefully, they can improve drafting, summarization, and first-pass research. The duty requires that you understand what you are using and that you verify what it produces. State bar guidance has converged on the same framing. The Florida Bar, in Ethics Opinion 24-1 issued January 19, 2024, advised that lawyers may use generative AI but must protect client confidentiality, guard against improper billing, and review AI output the way they would review the work of a nonlawyer assistant, verifying its accuracy and sufficiency. That supervisory analogy is a useful mental model: an AI tool is a fast, confident, sometimes unreliable assistant whose work always passes through a lawyer before it reaches a client or a court.

A workable competence baseline for a 3-to-30-attorney firm looks like this:

• Know the tool. Before using an AI product on client work, understand at a basic level what it does, how it handles the data you put into it, and whether your inputs are used to train the underlying model.
• Verify every output. Confirm that each cited case exists, says what the draft claims, and remains good law, using a primary source or a trusted database rather than the AI itself.
• Protect confidentiality. Do not enter client confidences into a consumer tool that may retain or train on them, and obtain specific informed consent rather than relying on boilerplate when client information is involved.
• Supervise the tool and the people using it. Apply your existing supervision duties to AI output as you would to a paralegal's, and make sure associates and staff know the verification rule is not optional.
• Bill honestly. Pass along efficiency gains rather than billing phantom hours, and do not charge clients for time spent learning the technology generally.
• Write it down. A short internal AI-use policy, even one page, gives the firm a defensible standard and a training tool for new hires.

None of these steps requires technical sophistication. They require the same judgment competence has always demanded, directed at a tool that is unusually good at sounding right while being wrong. That is precisely why the duty exists, and why keeping current with it is now part of basic competence rather than an optional specialty.

This is general information for lawyers and law-firm leaders, not legal advice, and it does not create an attorney-client relationship. The authorities are cited so you can read them yourself.