Telling Clients You Used AI: What NJ Attorneys Are Actually Required to Disclose Under RPC 1.4

Nobody talks about this one at CLE seminars. The discussions about AI in law practice tend to cluster around competence, confidentiality, and billing. But there's a quieter obligation sitting in plain sight inside RPC 1.4, and it deserves a direct look.

Here's the practical setup: you're a solo or small-firm attorney in New Jersey. You use an AI tool, say Casetext CoCounsel, Harvey, or even a well-prompted ChatGPT session, to draft a demand letter, summarize deposition transcripts, or prepare a motion brief. The work product looks good. You review it, edit it, sign off. You send it. Did you have to tell your client?

Most attorneys answer instinctively: "No, because I supervise the output and I'm responsible for the final product." That instinct is partially right, but it doesn't fully account for what RPC 1.4 actually demands.

What RPC 1.4 Requires, Read Carefully

RPC 1.4(b) requires an attorney to "explain a matter to the extent reasonably necessary to permit the client to make informed decisions regarding the representation." RPC 1.4(a) separately requires keeping the client "reasonably informed about the status of a matter."

Neither provision mentions technology. But the duty to inform isn't about technology for its own sake. It's about whether information is material to the client's ability to make decisions about the representation. That framing is where AI disclosure gets complicated.

Consider a client who, if asked, would object to their confidential facts being processed by a third-party AI vendor's servers. Or a client in a high-stakes custody dispute who wants to know whether the psychological evaluation summary you sent them was drafted by you or generated by a tool. The fact that you reviewed the output doesn't necessarily mean the client has the information they need to evaluate whether they're getting the representation they bargained for.

The New Jersey Supreme Court has not issued a formal ethics opinion specifically on AI disclosure as of mid-2025. The ABA issued Formal Opinion 512 in 2024, which addressed generative AI and concluded that disclosure may be required depending on the circumstances, particularly where confidentiality concerns are implicated. While ABA opinions aren't binding on NJ attorneys, the NJSBA routinely cites them as persuasive authority, and NJ's RPC 1.4 text tracks the ABA Model Rules closely enough that the reasoning applies.

The Three Situations Where Disclosure Is Probably Required

First, when AI tool use implicates your confidentiality obligations under RPC 1.6. If you're using a general-purpose AI tool that isn't covered by a proper data processing agreement, or whose terms of service allow training on your inputs, you may have already created a disclosure issue under RPC 1.6 separately. But under RPC 1.4, you have to ask whether the client would want to know their matter details were shared with that system. For most clients, yes.

Second, when the client has asked how the work is being done. This sounds obvious, but attorneys sometimes treat process questions as administrative rather than substantive. If a client asks "did you personally review all of these documents?" the accurate answer may involve disclosing that an AI tool handled the first-pass review. A technically accurate but incomplete answer could shade into the territory of RPC 8.4(c), which prohibits conduct involving dishonesty or misrepresentation.

Third, when the fee agreement doesn't address it. If your retainer agreement is silent on AI use and your client later learns you charged full hourly rates for work that was AI-assisted, the RPC 1.4 question converges with the RPC 1.5 question. The client's ability to make an informed decision about fees is compromised if they didn't know the method of production. This is especially true for flat-fee agreements where clients reasonably assumed they were paying for attorney time and judgment, not AI output plus review.

A Practical Approach for NJ Small Firms

The cleanest solution most NJ small firms aren't using is a brief AI disclosure paragraph in the engagement letter. It doesn't need to be elaborate. Something along these lines works: "Our firm may use AI-assisted tools to support legal research, drafting, and document review. All AI-generated work product is reviewed and approved by a licensed attorney before delivery. If you have questions about this practice, please ask."

This approach does several things at once. It satisfies the spirit of RPC 1.4(b) by giving clients the information they need to raise concerns or ask questions. It creates a record that disclosure was made. And it removes the awkward case-by-case decision about whether this particular matter, with this particular AI tool, crosses some undefinable threshold that requires a phone call.

Some attorneys worry that disclosing AI use will erode client confidence. The evidence in other professional services contexts runs the other direction: clients who are told about process efficiency tools and how they're supervised tend to trust the firm more, not less. Treat clients as adults who can handle accurate information about how their legal work gets done.

The practical next step: pull up your current engagement letter template this week and look for the gap. If there's nothing in it about how you use technology in the representation, you're leaving a disclosure decision to chance, and in New Jersey, that's a chance that isn't worth taking.