What NJ Solo Attorneys Get Wrong About Disclosing AI Use to Clients

There's a quiet disclosure problem spreading through New Jersey's solo and small-firm community — and it has nothing to do with hallucinations or data breaches. It's about the conversation attorneys are having (or not having) with their clients about AI.

Some attorneys are saying nothing. Others are burying a vague line in their engagement letters about "technology-assisted drafting." A few are going the opposite direction, offering unsolicited lectures on large language models that leave clients more confused than informed. Almost nobody is getting it right.

That gap has real professional consequences under NJ RPC 1.4, which governs client communication. And as AI becomes a routine part of legal practice, regulators are paying closer attention to whether clients are being kept "reasonably informed" about the methods being used on their behalf.

---

What RPC 1.4 Actually Requires (And What It Doesn't)

RPC 1.4 requires you to keep clients reasonably informed about the status of their matter and to promptly comply with reasonable requests for information. It also requires you to explain matters "to the extent reasonably necessary to permit the client to make informed decisions."

Notice what it does not say: it does not require you to disclose every tool you use. You don't file a notice every time you use Westlaw instead of Lexis. The relevant question isn't what software ran — it's whether the use of that tool affects the client's decision-making, risk exposure, or the reliability of the work product.

That distinction is where most NJ attorneys go wrong. The question you should be asking isn't "should I tell my client I used AI?" It's: "Does how I used AI change anything my client has a right to know?"

---

Three Scenarios Where Disclosure Is Genuinely Required

1. AI-Drafted Documents Given Directly to Third Parties If you used an AI tool to draft a contract, demand letter, or pleading, and that document goes to opposing counsel or a court with minimal attorney revision, your client arguably deserves to know the document's origin — especially if the quality or depth of analysis differs from what they'd reasonably expect from a fully attorney-drafted work product.

2. AI Used to Screen or Summarize Client-Provided Information Some attorneys are running client-uploaded documents through AI summarization tools before they read them. If that AI summary shapes your legal advice and you never read the underlying documents directly, that is a material fact affecting the quality of your counsel. Disclosure is appropriate.

3. AI That Generated Legal Research You Couldn't Independently Verify Citing AI-sourced research without independent verification violates RPC 3.3, but short of that line, if you're relying heavily on AI-generated research threads and the client's matter is high-stakes, reasonable transparency is good practice and defensible ethics.

---

What Good Disclosure Actually Looks Like

The goal of an AI disclosure isn't to confess — it's to demonstrate professional judgment. Consider adding a paragraph to your engagement letter that reads something like:

"This firm uses AI-assisted drafting and research tools as part of its standard workflow. All AI-generated work product is reviewed, edited, and approved by a licensed New Jersey attorney before use. You will always receive work that reflects independent professional judgment — not raw AI output. If you have questions about how these tools are used in your matter, please ask."

That language accomplishes four things: it informs, it reassures, it demonstrates competence under RPC 1.1, and it opens a dialogue. It does not bury the client in technical detail or imply that AI is doing the lawyering.

---

The Mistake of Saying Too Much

A counterintuitive risk: attorneys who over-explain AI to clients — particularly non-sophisticated individual clients — can inadvertently undermine their own authority. Paragraphs of caveats about "potential hallucinations" and "probabilistic language models" read less like transparency and more like an attorney pre-excusing bad work.

Disclosure should be calibrated to what the client needs to make informed decisions — not what makes you feel ethically covered. If your disclosure language is longer than your fee section, it's almost certainly too long.

---

The Practical Takeaway

New Jersey attorneys don't need a separate AI disclosure policy — they need to integrate AI communication into the existing framework of RPC 1.4. That means:

• Engagement letter: One clear, reassuring paragraph on AI use and attorney oversight.
• Matter-specific communication: A brief note when AI played a materially significant role in a specific deliverable.
• Client questions: A standing offer to discuss your process, which few clients will take you up on but many will appreciate having.

The Bar isn't currently mandating specific AI disclosure language, but guidance from the ACPE and analogous opinions from states like California and New York signal the direction of travel. Getting your disclosure posture right now — when it's a best practice — is far easier than retrofitting it after it becomes a rule.

Your clients don't need to understand AI. They need to trust you. A well-crafted disclosure statement builds that trust rather than testing it.