Telling Clients You Used AI — Does NJ RPC 1.4 Actually Require It?

Most of the AI ethics conversation in New Jersey has orbited confidentiality and competence. That's understandable — RPC 1.6 and RPC 1.1 are high-stakes, and bar guidance has pointed squarely at them. But there's a quieter rule sitting in the background that solo and small firm attorneys are underestimating: RPC 1.4, which governs communication with clients.

The question is deceptively simple: if AI drafted your motion, summarized your deposition transcripts, or flagged the key issues in opposing counsel's contract — do you have to tell your client?

The answer isn't a clean yes or no. But the analysis matters, and getting it wrong creates real exposure.

---

What RPC 1.4 Actually Says

New Jersey RPC 1.4(b) requires that a lawyer "explain a matter to the extent reasonably necessary to permit the client to make informed decisions regarding the representation." RPC 1.4(a) adds that a lawyer must keep the client "reasonably informed about the status of a matter."

Notice what those provisions don't say: they don't say you must narrate your workflow. You're not required to tell a client you used Westlaw, dictated into Dragon, or outsourced document formatting to a contract paralegal in another state. The method of your work product isn't typically a communication obligation.

So at first glance, AI use looks like just another tool — no disclosure required.

But here's where it gets complicated.

---

When AI Changes the "Informed Decision" Calculus

RPC 1.4's "informed decisions" language is the operative hook. Clients have the right to make meaningful choices about their representation — including decisions about cost, strategy, and risk. If AI meaningfully affects any of those dimensions, disclosure edges from optional toward obligatory.

Consider three scenarios common in NJ small firm practice:

Scenario 1 — The Fee Question. You used an AI drafting tool to produce a 15-page contract in 40 minutes instead of four hours. You've already billed at a reduced rate under a value-billing arrangement. The client doesn't need to know the mechanism — but if they're paying hourly and the matter was substantially completed by AI, RPC 1.4 intersects directly with RPC 1.5's fee reasonableness standard. The communication obligation and the billing obligation converge.

Scenario 2 — The Accuracy Risk. You used a general-purpose LLM (not a purpose-built legal AI) to summarize a 200-page discovery production. The tool missed three documents flagged as potentially privileged by opposing counsel. Your client later faces a privilege waiver argument they didn't know was possible. Would disclosure of the AI's role — and its limitations — have let your client make a more informed decision about the review methodology? Arguably yes. That's a RPC 1.4 conversation you should have had.

Scenario 3 — The Client's Own Views. Some clients — particularly in sensitive family law, immigration, or criminal defense matters — have strong feelings about their personal information being processed by third-party AI systems. Even if your vendor relationship is airtight from a RPC 1.6 standpoint (BAAs signed, data residency confirmed), your client may reasonably want to weigh in on whether that tradeoff is acceptable. RPC 1.4 exists precisely to preserve that autonomy.

---

What Proactive Disclosure Actually Looks Like

You don't need a three-page AI addendum attached to every engagement letter. What you need is proportionate, upfront communication — and the right moment to have it is the retainer conversation, not after a problem surfaces.

A practical approach for NJ solo attorneys:

1. Add one paragraph to your engagement letter stating that the firm may use AI-assisted tools to improve efficiency and accuracy in research, drafting, and document review, and that these tools are selected with client confidentiality in mind.
2. Flag substantive AI use in your file notes. If AI did meaningful analytical work on a matter — not just formatting, but analysis — document that you reviewed and verified the output. This protects you under RPC 5.3 supervision obligations simultaneously.
3. Have the conversation when accuracy risk is elevated. If you're using AI to summarize voluminous records, do a quick review of the tool's limitations with your client before relying on its output in a consequential filing. Two minutes of candor now prevents a bar complaint later.
4. Don't over-disclose in a way that undermines confidence. RPC 1.4 is not a confession booth. Clients hire you for judgment. Disclosure should convey competent, deliberate use of tools — not anxiety about them.

---

The Bigger Picture for NJ Small Firms

The New Jersey Supreme Court and the ACPE haven't yet issued specific guidance tying RPC 1.4 to AI disclosure the way other jurisdictions have begun to. But the California State Bar's 2024 AI guidance explicitly flagged client communication obligations, and the ABA's Formal Opinion 512 (2024) on generative AI touched on the informed consent dimension. New Jersey practitioners would be naive to assume the issue won't arrive here.

Getting ahead of it now — with a simple, clear engagement letter paragraph and a thoughtful internal policy — costs you almost nothing. Getting behind it after a client complaint costs considerably more.

RPC 1.4 was written for a world of telephone calls and letter campaigns. It applies just as cleanly to the age of AI-assisted practice. The duty to communicate hasn't changed — only the conversations worth having.