Disclosing AI Use to Clients: The NJ Ethics Conversation Most Small Firms Are Quietly Avoiding

There's a quiet assumption spreading through NJ small firms: if the AI-assisted work product is good, there's no reason to mention how it was made.

That assumption deserves a hard look.

As AI use becomes routine — not just for legal research but for drafting demand letters, intake summaries, contract redlines, and even settlement analyses — the question of whether clients have a right to know about it is no longer theoretical. New Jersey's Rules of Professional Conduct don't yet contain the word "artificial intelligence." But several existing rules, read together, create disclosure obligations that most solo practitioners are only dimly aware of.

Why Silence Isn't Neutral

Let's start with the practical reality. When a client hires a solo attorney in Trenton or a three-lawyer firm in Hackensack, they're entering a relationship built on trust and reasonable expectations. If they later discover that the brief filed on their behalf was substantially drafted by a generative AI tool — one that charged them full hourly rates while doing so — and no one told them, how would they react?

That's not a hypothetical. It's increasingly the kind of complaint that lands in front of the NJ Office of Attorney Ethics.

The relevant rules aren't obscure. RPC 1.4(b) requires attorneys to explain matters "to the extent reasonably necessary to permit the client to make informed decisions regarding the representation." If AI is shaping the substance of work product or affecting how efficiently work gets done, clients arguably have an interest in that information — especially if it touches billing practices. And if AI materially reduces the time required for a task, RPC 1.5's reasonableness standard for fees starts asking harder questions about what exactly the client is paying for.

There's also a RPC 1.6 dimension that firms consistently underestimate. The moment a client's facts — their dispute, their contract terms, their personal circumstances — flow into a cloud-based AI tool, data handling becomes a confidentiality question. Disclosure isn't just about good faith. It's about whether the client has given informed consent to how their information is being processed.

What the NJ Bar Has (and Hasn't) Said

As of mid-2025, the New Jersey State Bar Association has not issued a formal ethics opinion specifically addressing generative AI disclosure. But the NJSBA's guidance on technology competence, combined with the ABA's Formal Opinion 512 (which NJ practitioners should be monitoring even if not bound by it), points in a clear direction: attorneys have an obligation to understand the tools they use, and that understanding should be transparent to clients when the tools touch their matter.

Several state bars — including Florida and California — have moved toward affirmative disclosure requirements. NJ hasn't gone that far yet, but "not required yet" is a risky basis for a client relationship strategy.

What a Proper Disclosure Actually Looks Like

Good AI disclosure to clients doesn't require a lengthy rider or a lecture. It should accomplish three things:

1. Inform the client that AI tools may assist in their matter (draft documents, research, summarize records, etc.)
2. Reassure them that all AI output is reviewed and verified by a licensed attorney before being used
3. Explain how their data is protected — specifically that you use tools compliant with your confidentiality obligations under NJ RPC 1.6

Here's a plain-language example that can be incorporated into a standard engagement letter or client intake packet:

"Our firm uses AI-assisted tools to support legal research, document drafting, and case analysis. All AI-generated work product is reviewed and approved by a licensed NJ attorney before being used in your matter. We only use tools that meet our confidentiality obligations, and your information is never used to train third-party AI models. You are always welcome to ask which tools were used in connection with your representation."

That's 68 words. It takes ten seconds to read. And it meaningfully reduces your exposure under RPC 1.4, RPC 1.5, and RPC 1.6 simultaneously.

The Billing Dimension No One Wants to Talk About

Here's the edge most attorneys are avoiding: if AI cuts a four-hour drafting task to forty-five minutes, are you billing four hours? One hour? A flat fee that was calibrated before AI existed?

The RPC 1.5 question here is genuinely unsettled, but the ethical weight is real. Some practitioners are handling this by converting AI-assisted work to flat-fee arrangements. Others are building a "technology fee" line item into their engagement letters — a practice that only works if it's disclosed upfront. Whatever your approach, your engagement letter is where it needs to live. Silence in the contract is exposure in the complaint.

Start Before You Have a Problem

The NJ attorneys who will be best positioned when formal guidance does arrive — and it will — are the ones who built disclosure habits into their intake process now. That means updating engagement letters, having a brief conversation with new clients about your practice's use of AI tools, and documenting that the conversation happened.

This isn't about being defensive. It's about being the kind of practitioner clients trust precisely because you told them things other attorneys assumed didn't need saying.