ABA Model Rule 1.1 and AI Notetakers: What the Duty of Technological Competence Requires
When the American Bar Association added Comment 8 to Model Rule 1.1 in 2012, the change was short — twenty-one words that grafted a duty of technological competence onto the general duty of competent representation. More than a decade later, that comment has been adopted in some form by roughly 40 states, and it is the doctrinal hook nearly every ethics opinion on generative AI hangs from. The 2024 release of ABA Formal Opinion 512 and, more recently, NYC Bar Formal Opinion 2025-6 made explicit what many practitioners already suspected: turning on an AI notetaker without understanding how it processes client audio is not a neutral act. It is a competence question.
This article walks through what Rule 1.1 actually asks of lawyers who use AI transcription and summarization tools, where the risk sits, and what a defensible evaluation looks like in practice.
What Model Rule 1.1 and Comment 8 Actually Say
ABA Model Rule 1.1 requires "the legal knowledge, skill, thoroughness and preparation reasonably necessary for the representation." Comment 8 clarifies that maintaining competence includes keeping abreast of "the benefits and risks associated with relevant technology." The comment does not require lawyers to become engineers. It requires them to understand enough to make informed decisions — and, where they cannot, to consult someone who can.
For AI notetakers, the "benefits and risks" phrase does the heavy lifting. Benefits are obvious: fewer missed details, better client service, less time on post-meeting summaries. Risks are less obvious and are almost entirely a function of how the tool is architected. Two notetakers can look identical in the UI and have radically different privilege, confidentiality, and third-party-disclosure profiles depending on where the audio goes after the microphone captures it.
The Overlap with Rule 1.6 Confidentiality
Competence under Rule 1.1 cannot be separated from confidentiality under ABA Model Rule 1.6. Rule 1.6(c) requires lawyers to "make reasonable efforts to prevent the inadvertent or unauthorized disclosure of, or unauthorized access to, information relating to the representation." Comment 18 lists factors that inform reasonableness: the sensitivity of the information, the likelihood of disclosure absent additional safeguards, the cost of employing safeguards, and the difficulty of implementing them.
An AI notetaker that streams audio to a cloud service, stores transcripts on vendor infrastructure, and reserves rights to use content for model training presents a very different Rule 1.6 calculus than one that processes audio on the lawyer's device and never transmits it. A competent lawyer is expected to know which category their tool falls into. "I didn't realize the vendor kept a copy" is not a defense — it is the failure Rule 1.1 was amended to address.
What Formal Opinion 512 Requires
ABA Formal Opinion 512, issued in July 2024, addresses generative AI directly. Its central framing is that existing rules already govern AI use — no new rules are needed, but existing ones must be applied thoughtfully. The opinion identifies several concrete obligations that map cleanly onto notetaker deployments:
- Understand the tool. Lawyers must have a "reasonable understanding" of how the AI handles inputs and outputs, including where data is stored and whether it is used for training.
- Protect confidential information. Inputting client information into a self-learning tool without informed client consent can violate Rule 1.6.
- Communicate with the client. Rule 1.4 may require disclosure of AI use, particularly where the tool materially affects the representation or where the client would reasonably want to know.
- Supervise appropriately. Rules 5.1 and 5.3 apply to AI as a non-lawyer assistant; firm leadership must have policies in place.
Notetakers are squarely within this framework. Meeting audio with a client is confidential information relating to the representation. The moment that audio leaves the lawyer's device, Rule 1.6 is implicated. The competence duty under Rule 1.1 is what obligates the lawyer to know, in advance, whether the audio leaves at all.
NYC Bar Formal Opinion 2025-6
The New York City Bar Association's Professional Ethics Committee issued Formal Opinion 2025-6 in December 2025, focused specifically on generative AI in law practice. It reinforces the ABA framework and adds practical texture, emphasizing that lawyers must scrutinize vendor terms of service, data retention practices, and training-data policies. The opinion treats vague or shifting vendor representations as a red flag: if a lawyer cannot determine from the contract and documentation what happens to client data, the competence and confidentiality analysis has not been completed.
This matters for notetakers because vendor practices vary widely. Some tools default to using conversation content to improve models unless the user opts out. Some retain transcripts indefinitely. Some route audio through multiple subprocessors located in jurisdictions with different data-protection regimes. A lawyer who has not read the current terms — not last year's, the current ones — has not satisfied the inquiry Rule 1.1 requires.
The Third-Party Disclosure Problem
The competence analysis takes on additional weight in light of recent case law on AI and privilege. In US v. Heppner (S.D.N.Y. Feb 2026), Judge Rakoff held that a litigant's chats with a public AI platform were not privileged, drawing on the third-party doctrine: content voluntarily disclosed to a third party generally loses its privileged character. The court analogized the AI provider to any other third-party recipient of communications.
Applied to notetakers, the analysis is uncomfortable. When a lawyer records a privileged client conversation and transmits the audio to a cloud vendor for transcription and summarization, that vendor becomes a recipient. Whether privilege is preserved will depend on whether the vendor qualifies as an agent of the lawyer for privilege purposes (a fact-intensive inquiry that turns on the terms of the engagement and the necessity of the service), and on whether waiver has occurred through the disclosure itself.
The Heppner court did not rule on notetaker vendors specifically, and there are meaningful distinctions between a public chatbot and a contracted transcription service. But the direction of judicial reasoning is a data point Rule 1.1 requires lawyers to weigh. See also Brewer v. Otter.ai (2025), a putative class action raising wiretap and privacy claims against a major notetaker vendor over recording and retention practices — the kind of litigation that puts vendor architecture squarely on the record.
Vendor Diligence: What a Competent Evaluation Looks Like
A defensible Rule 1.1 evaluation of an AI notetaker is not a checkbox exercise. It is a written record of what the lawyer asked, what the vendor answered, and what the lawyer decided. The following table summarizes the categories that should be documented before a tool touches a client conversation.
| Evaluation Category | Question to Answer | Why It Matters Under Rule 1.1 |
|---|---|---|
| Data flow | Where does audio go after capture? Device only, or transmitted to a server? | Determines whether third-party disclosure has occurred at all. |
| Storage | Where are transcripts and summaries stored, and for how long? | Longer retention on vendor infrastructure expands the subpoena and breach surface. |
| Subprocessors | Which third parties (transcription APIs, model providers, cloud hosts) touch the data? | Each subprocessor is a separate disclosure to evaluate. |
| Training use | Is client content used to train models, by default or opt-in? | Formal Opinion 512 flags this as a core Rule 1.6 issue. |
| Contracts | Does the vendor sign a DPA and BAA where applicable? NDA on request? | Written protections are the backbone of a reasonable-efforts defense. |
| Consent workflow | How does the tool support recording consent in one-party and two-party jurisdictions? | Recording-consent statutes are strict-liability in several states. |
| Access controls | Who at the vendor can access transcripts? What audit logs exist? | Insider access is a documented breach vector. |
| Deletion | Can the lawyer delete audio and transcripts on demand, and is deletion verifiable? | Retention limits reduce exposure across the life of the matter. |
Common Failure Modes
Three patterns repeat in the ethics literature and in reported incidents:
Bot-in-the-meeting deployments. Many popular notetakers work by sending a bot to join a Zoom, Teams, or Meet call as an additional participant. The bot receives the audio stream and forwards it to vendor infrastructure. This architecture makes the vendor a visible third party to every participant on the call and creates a durable record on vendor servers. It also raises consent questions in two-party jurisdictions because the bot's presence is often the only notice given.
Default training opt-in. Some tools use conversation content to improve models unless the user actively disables the setting. A lawyer who has not audited settings may be feeding client communications into a training pipeline without knowing it. Formal Opinion 512 treats this as a serious Rule 1.6 problem.
Silent scope expansion. Vendors change terms of service. A tool that did not use content for training last year may this year. Rule 1.1 competence is not a one-time evaluation — it is an ongoing obligation. See West Technology Group v. Sundstrom (D. Conn. 2024) for a reminder that vendor conduct and contract terms are frequently litigated and that lawyers are expected to track them.
Client Communication Under Rule 1.4
Formal Opinion 512 signals that AI use may trigger Rule 1.4 disclosure obligations. Whether a specific notetaker requires client notice depends on several factors: the sensitivity of the matter, the client's likely expectations, whether the tool transmits data to third parties, and whether the tool materially affects the representation (for example, by generating summaries the lawyer relies on for strategy decisions).
A reasonable practice for cloud-based notetakers is to disclose their use in the engagement letter and obtain written consent. For on-device tools that do not transmit data, disclosure may still be prudent as a matter of client trust, but the Rule 1.6 disclosure risk is materially different because there has been no third-party transmission to disclose.
For a deeper look at the consent architecture side of this, see our guide to recording consent across two-party states and our overview of privilege risk with AI notetakers.
Supervision Under Rules 5.1 and 5.3
Rule 1.1 does not sit alone. Partners and managing lawyers have supervisory duties under Rules 5.1 and 5.3 that require them to ensure firm-wide compliance. For AI notetakers, this typically means a written policy addressing which tools are approved, what settings are required, what matters are excluded (for example, matters involving especially sensitive information), and how consent is obtained and logged.
A firm that allows attorneys to install any notetaker they find in the App Store has not satisfied Rule 5.1. A firm that has evaluated one or two tools, documented the evaluation, and issued a written policy has.
How Basil approaches this
Basil is built to reduce the surface area a Rule 1.1 evaluation has to cover. Audio, transcription, and summarization all run on-device using the Apple Neural Engine. Nothing is uploaded, there is no server on the audio path, and Basil never receives user data — so the third-party disclosure analysis that drove Heppner and animates Formal Opinion 512 looks materially different. There is no bot that joins Zoom, Teams, or Meet calls; capture happens locally in Computer mode on macOS. We sign DPAs and NDAs on request.
The Basil for Law edition — with privilege attestation, consent logging, matter organization, and Privileged & Confidential labeling — launches in August 2026 at $19.99/month or $199.99/year for solo attorneys, with a 3-day trial on the monthly plan and a 7-day trial on the annual plan. The general Basil app, with a free 60-minutes-per-month tier, is available today. None of this eliminates the lawyer's independent obligation under Rule 1.1 to evaluate the tool, but it is designed so that the evaluation is short and the answer is architectural.
This article is for information only and is not legal advice.
Frequently asked questions
Does Model Rule 1.1 require me to understand how my AI notetaker works technically?
Comment 8 requires a reasonable understanding of the benefits and risks — not engineering-level knowledge. For a notetaker, that means knowing where audio goes after capture, whether transcripts are stored on vendor infrastructure, whether content is used for training, and what contractual protections apply. If you cannot answer those questions from the vendor's documentation, the competence inquiry is not complete.
Is using a cloud-based AI notetaker a violation of Rule 1.6?
Not automatically. Rule 1.6(c) requires reasonable efforts to prevent unauthorized disclosure, and reasonableness is a multi-factor analysis. A cloud notetaker with a signed DPA, no training use of client content, defined retention, and appropriate access controls may satisfy the standard. A tool with vague terms and default training opt-in likely does not. Formal Opinion 512 makes the analysis explicit.
Do I have to tell clients I use an AI notetaker?
Formal Opinion 512 indicates Rule 1.4 disclosure may be required, particularly where the tool transmits data to third parties or materially affects the representation. For cloud tools, written disclosure in the engagement letter and documented consent is a common approach. For on-device tools that do not transmit data, disclosure remains prudent for trust reasons but the underlying Rule 1.6 posture is different.
What did US v. Heppner actually hold about AI and privilege?
In Heppner (S.D.N.Y. Feb 2026), Judge Rakoff held that a litigant's chats with a public AI platform were not privileged, applying third-party doctrine reasoning. The case did not address notetaker vendors specifically, but the analytical direction is a data point competent lawyers should weigh when evaluating tools that transmit client communications to third parties.
Is a bot that joins my Zoom calls a Rule 1.1 problem?
Bot-based notetakers introduce a visible third party to the call and create a durable record on vendor infrastructure. That does not automatically violate any rule, but it expands the disclosure, retention, and consent analysis the lawyer must perform. In two-party consent jurisdictions, the bot's presence may also be the only notice given to other participants, which is often insufficient on its own.
How often do I need to re-evaluate my notetaker under Rule 1.1?
Competence is ongoing. Vendors change terms of service, add subprocessors, and update training-data policies. A reasonable practice is to review vendor terms at least annually and whenever the vendor announces material changes, and to document the review. NYC Bar Formal Opinion 2025-6 treats stale diligence as a red flag.
Keep client conversations on your device
Basil transcribes and summarizes entirely on-device — no cloud, no bot, no server to subpoena. See Basil for Law → · Legal-tool reviews →
This article is for information only and is not legal advice.