After the ABA's Formal Opinion 512, Here's What NJ Small Firms Must Do Differently Right Now

In July 2024, the ABA Standing Committee on Ethics and Professional Responsibility released Formal Opinion 512 — its first comprehensive guidance on lawyers' use of generative AI tools. If you practice in New Jersey and haven't yet read it, you're not alone. But "haven't read it" and "not affected by it" are two very different things.

NJ's Rules of Professional Conduct don't exist in a vacuum. When the ABA issues a formal opinion of this weight, New Jersey disciplinary panels take notice. The ACPE (Advisory Committee on Professional Ethics) has historically cited ABA opinions as persuasive authority, and Opinion 512 directly implicates at least five RPCs that New Jersey has adopted in substantially the same form. Here's what the opinion says, what it means for your practice, and the three concrete steps you should take before opening any AI tool on your next client matter.

---

What ABA Formal Opinion 512 Actually Says

The opinion is structured around four core duties:

1. Competence (RPC 1.1): Lawyers must understand the benefits and risks of any AI tool they use — not just generally, but specifically enough to evaluate output quality. Delegating work to an AI assistant you don't understand is, per the ABA, an ethical failure.

2. Confidentiality (RPC 1.6): Before inputting any client information into a generative AI tool, you must conduct a reasonable inquiry into how that tool stores, transmits, and potentially trains on your data. Default consumer settings on most major platforms do not satisfy this standard.

3. Supervision (RPCs 5.1 / 5.3): If a non-lawyer on your team — a paralegal, legal assistant, or contract researcher — uses AI to produce work product, the supervising attorney remains fully responsible for reviewing that output with the same diligence they would apply to any delegated task.

4. Candor and Fees (RPCs 3.3 / 1.5): AI-generated legal citations must be independently verified. And if AI substantially reduces the time required to complete a task, billing the full pre-AI hourly rate for that task may raise fee reasonableness questions.

None of this is radical. But here's the challenge for NJ small firms: the opinion assumes you have systems in place to meet these duties consistently. Most solo practitioners and two- to four-attorney firms are running AI tools ad hoc — no written policy, no vendor review checklist, no staff guidance. That gap is precisely where ethical exposure lives.

---

Where New Jersey Specifically Sharpens the Obligation

New Jersey's version of RPC 1.1 includes an ongoing duty of technological competence that mirrors the ABA Model Rules' 2012 amendment — and NJ courts have been willing to engage with it seriously. More pointedly, NJ RPC 1.4 (communication) adds an obligation that Opinion 512 only gestures at: if your use of AI materially affects the nature of the services you're providing to a client, there is a credible argument that the client is entitled to know.

This isn't the disclosure debate from first principles — it's narrower and more practical. If you're using AI to generate the first draft of a brief that you previously would have researched and written yourself over eight hours, and you're billing for that brief, your client's understanding of what they're paying for has shifted. NJ's communication rules are sufficiently broad to reach this scenario, even if no formal opinion has yet said so explicitly.

---

Three Operational Steps Before You Open That AI Tool

1. Run a thirty-minute vendor review before first use — and document it. Pull up the terms of service and privacy policy for any AI tool your firm uses. Answer three questions in writing: (a) Does the vendor use my inputs to train its models? (b) Where is data stored, and is it subject to U.S. law? (c) Is there a data processing agreement or enterprise tier available? If you can't answer all three, you're not ready to use it on client matters. Keep a one-page memo in your file — it's your paper trail if a complaint is ever filed.

2. Create a single-page AI output review protocol for your office. This doesn't need to be a 20-page policy. It needs to answer: who on the team is authorized to use AI tools, which tools are approved, and what verification step must occur before AI-generated text enters any court filing or client deliverable. A paralegal who knows the rule is not a supervisory failure. A paralegal who doesn't is.

3. Revisit your fee agreements with one specific question in mind. Does your current retainer agreement contemplate AI-assisted work? If you bill hourly, does it explain that you may use technology to improve efficiency, and that the client benefits from the time savings while still paying for your professional judgment? A single paragraph addressing this now is far cleaner than an uncomfortable fee dispute later — or a grievance alleging you billed for work an AI performed in minutes.

---

Opinion 512 isn't the finish line on AI ethics guidance — it's the starting line. State bars, including New Jersey's, will issue their own formal opinions in the months and years ahead. Firms that have already built the operational scaffolding to meet these standards will adapt easily. Firms that are still winging it will find themselves retrofitting a policy under pressure.

The time to build that scaffolding is before the complaint, not after.