ABA Rule 5.3 and Your Legal Tech Vendors: Reasonable Supervision

Key takeaways

Since 2012, Model Rule 5.3 covers nonlawyer assistance, including outside services such as cloud storage, software vendors and consultants.
The standard is reasonable efforts scaled to the risk, not a fixed audit. Comment [3] lists the factors that set how much effort is reasonable.
ABA Formal Opinion 512 (2024) applies the same vendor diligence to generative AI providers, including whether they retain your data or claim rights to it.
Write the answers into the contract, and prefer designs where the data, access and change history stay in an environment your firm controls.

Your vendors are part of your practice

A 30-attorney firm might rely on a practice management platform, a document management system, a phone system, an e-signature service, a managed IT provider and a consultant who builds integrations. Each one touches client information. Under the ABA Model Rules of Professional Conduct, the people behind those services are nonlawyers assisting the firm, and the firm’s lawyers have a duty to supervise them.

This article explains what the Model Rules and the main ABA ethics opinions say about that duty, then turns it into a checklist for managing partners and firm administrators. It covers the ABA Model Rules only. Your state’s rules and ethics opinions may differ, so check your own jurisdiction.

This is general information, not legal advice. Circa Logic is not a law firm. For questions about your obligations in a specific situation, talk to your firm’s ethics counsel or your state bar’s ethics hotline.

What Rule 5.3 actually says

Model Rule 5.3 has three parts. Partners and lawyers with comparable managerial authority must make reasonable efforts to ensure the firm has measures in place giving reasonable assurance that nonlawyers’ conduct is compatible with the lawyers’ professional obligations. A lawyer with direct supervisory authority over a nonlawyer must make reasonable efforts to the same end. And a lawyer is responsible for a nonlawyer’s conduct that would violate the rules if the lawyer orders or ratifies it, or, holding managerial or supervisory authority, knows of it in time to avoid or limit the consequences and fails to take reasonable remedial action.

In 2012, as part of the Ethics 20/20 amendments, the ABA changed the rule’s title from “Nonlawyer Assistants” to “Nonlawyer Assistance.” The change was small in wording and large in reach. The rule is no longer read as being only about secretaries and paralegals. It covers services, including services from companies outside the firm.

The comments make that explicit. Comment [3] says “a lawyer may use nonlawyers outside the firm to assist the lawyer in rendering legal services to the client,” and its examples include a document management company and “an Internet-based service to store client information.” When using such services, a lawyer “must make reasonable efforts to ensure that the services are provided in a manner that is compatible with the lawyer’s professional obligations.” How much effort is reasonable depends on the provider’s education, experience and reputation, the nature of the services, the terms protecting client information, and the legal and ethical environment where the work is done. Comment [4] adds that where a client chooses the provider, the lawyer ordinarily should agree with the client on who monitors it.

The rule does not require a firm to audit every vendor or read source code. The standard is reasonable efforts, scaled to the risk.

The rules and opinions that sit beside it

Rule 5.3 rarely works alone. Several other provisions shape what reasonable supervision of a technology vendor looks like.

ABA formal opinions are not binding on their own. States adopt, modify or depart from them, and many have their own opinions on cloud storage and AI.

Rule 1.1, Comment [8] (competence). To stay competent, a lawyer should keep abreast of changes in the law and its practice, “including the benefits and risks associated with relevant technology.” You do not need to be a technologist, but you need enough understanding to ask the right questions, or help from someone who has it.
Rule 1.6(c) (confidentiality). A lawyer “shall make reasonable efforts to prevent the inadvertent or unauthorized disclosure of, or unauthorized access to, information relating to the representation of a client.” Comment [18] lists factors for judging reasonableness, including the sensitivity of the information, the likelihood of disclosure without added safeguards, and the cost and difficulty of those safeguards.
ABA Formal Opinion 08-451 (2008, outsourcing). A lawyer may outsource legal or nonlegal support services provided the lawyer “remains ultimately responsible for rendering competent legal services to the client.”
ABA Formal Opinion 477R (2017, securing client communications). Favors a fact-based analysis over a fixed list of required security measures, and calls for due diligence on vendors that provide communication technology.
ABA Formal Opinion 498 (2021, virtual practice). The duty to supervise nonlawyers extends to those inside and outside the firm, and lawyers should review the terms of service of their software to see whether confidentiality is protected.
ABA Formal Opinion 512 (July 2024, generative AI). Lists diligence points from earlier opinions, such as reference checks, vendor credentials, security policies, hiring practices and confidentiality agreements, and says “these concepts also apply to GAI providers and tools.”

What Opinion 512 adds for AI tools

Opinion 512 is worth reading in full if any of your vendors use AI on client data, as many practice management, document and phone products now do. As a baseline, lawyers should read and understand the terms of use, privacy policy and related terms of any generative AI tool they use, or consult someone who has. The opinion also concludes that putting client information into a self-learning tool that could reveal it to others requires the client’s informed consent, and that engagement-letter boilerplate is not enough.

Drawing on earlier cloud computing and outsourcing opinions, Opinion 512 suggests lawyers should:

confirm the tool is configured to preserve confidentiality and security, that the obligation is enforceable, and that the firm will be notified of a breach or of legal process seeking client information;
investigate the tool’s reliability, security measures and policies, including limits on the provider’s liability;
determine whether the tool retains information the firm submits, before and after the service ends, or claims proprietary rights in it;
understand that the provider’s servers can fail and may be an attractive target for cyberattacks.

A vendor due-diligence checklist

Here is how we turn those principles into questions. Ask them of every vendor or consultant that touches client data, including us, and keep the answers on file. A written record of what you asked, what you were told and what you decided is the clearest way to show what the firm did.

Where the data lives and who owns the environment. Is client data stored and processed in the vendor’s cloud, a subcontractor’s, or your firm’s own tenant? Who holds the administrator rights?
Who can access it, and how access is granted and removed. Which of the vendor’s people can see client data? Is access named, protected by multi-factor authentication and logged? Who at your firm can remove it, and how quickly?
Login and admin surfaces. Is there a web portal, admin console or support back door? How is it protected, and could someone use it to change the system without your firm knowing?
Change control. How do changes reach production? Can your firm see each change, who made it and why, and require its own approval before it goes live?
AI training and retention. Does the vendor, or any model provider it uses, retain prompts, documents or outputs? Are they used to train or improve models? Is that prohibited in the contract, not only on a policy page that can change?
Incident notification. How quickly will you be told about a suspected incident, through what channel and with what information? Will you also be told about subpoenas or other demands for your data?
Insurance. Does the vendor carry technology errors and omissions and cyber coverage? Ask for a certificate.
End of engagement. What happens to your data, the accounts and the system when the relationship ends? Is return and deletion confirmed in writing?
Subcontractors. Who else processes your data for the vendor? Does adding one need your approval, and are they bound by equally protective terms?

Put it in the contract

Answers on a sales call are not commitments. Most firms put the important ones into a master services agreement plus a data protection or confidentiality addendum, reviewed by their own counsel. The aim is a record that the firm considered the risks Comment [3] describes and dealt with them. Terms worth looking for:

use of client data limited to performing the services under your instructions, with no sale, no secondary use and no AI training;
all firm data treated as confidential and potentially privileged, whether or not it is marked;
a written list of the access and permissions the vendor holds, kept current and removed at the end;
a specific incident notice window and a duty to cooperate, with notifications to clients and regulators left under the firm’s control;
notice of subpoenas and other legal demands for firm data, where lawful;
prior written approval of subcontractors, on terms at least as protective;
a reasonable right for the firm to send security questionnaires and review compliance;
return or deletion of data at the end of the engagement, confirmed in writing;
stated insurance requirements.

Why in-tenant architecture makes supervision easier

Much of that checklist gets simpler when a vendor’s work runs inside an environment your firm already controls. If an integration is deployed into your own Microsoft 365 and Azure tenant, the data stays where it already is, under the identity, multi-factor authentication, conditional access, retention and logging policies your firm already applies to email and documents.

Access becomes something your firm issues rather than something the vendor reports. Consultants sign in with accounts your firm creates, and your IT team can remove them in minutes. The automation runs under its own identity with only the permissions it needs, listed in writing.

Change control becomes visible. When the code lives in a repository your firm owns, every change is recorded with who made it and when, and your firm can require a named approver before anything goes live. That is the kind of measure Rule 5.3(a) asks managers to have in place.

The end of an engagement stops being a negotiation. The resources, logs and code are already yours. Removing the vendor means disabling its accounts, and the system keeps running.

This is how we build at Circa Logic: rules first, AI only where useful and always reviewed by a person, and no portal to log into. It does not remove your duty to supervise. It makes that duty easier to carry out and to show.

Keep it proportionate

A vendor that stores every client file deserves more scrutiny than one that books your conference room. A practical routine for a firm of 10 to 75 attorneys:

keep an inventory of every vendor that touches client data, with a named owner at the firm for each;
run the checklist before signing, and again when a vendor adds AI features or changes its terms;
review vendor accounts and permissions at least once a year and whenever people leave;
write down the firm’s policy on permissible AI use, which Opinion 512 says managerial lawyers must establish, and train staff on it;
check your state’s rules and ethics opinions, which may add requirements beyond the ABA model.