Ghostwriting Legal Briefs With AI — Does NJ RPC 8.4 Have Anything to Say About It?

There's a conversation happening in small NJ firms right now that almost nobody is labeling correctly.

An attorney feeds a fact pattern into an AI tool, gets back a polished argument, cleans it up, and files it. They call that "using AI to draft." What they haven't asked — and what their malpractice carrier probably will someday — is whether they've verified that work carefully enough to put their name on it without reservation.

That question isn't just about competence under RPC 1.1, which this blog has covered. It has a quieter, more serious cousin: RPC 8.4, New Jersey's professional misconduct rule. And it deserves a direct look.

---

What RPC 8.4 Actually Covers

Most attorneys think of RPC 8.4 as the rule that catches the dramatic stuff — fraud, dishonesty, crimes involving moral turpitude. And it does cover all of that. But subsections (c) and (d) are broader than they're often given credit for:

• RPC 8.4(c) prohibits conduct involving dishonesty, fraud, deceit, or misrepresentation.
• RPC 8.4(d) prohibits conduct prejudicial to the administration of justice.

Neither requires a crime. Neither requires intentional wrongdoing. And in the context of AI-assisted brief writing, both are worth thinking about carefully.

---

The Specific Scenario NJ Solo Attorneys Should Be Stress-Testing

You use an AI tool to draft a section of a brief. The AI attributes a legal standard to a case — let's say it cites State v. [Plausible-Sounding Name] for a proposition about the reasonable expectation of privacy in New Jersey. You Westlaw it quickly, it doesn't come up immediately, and you move on. The brief gets filed.

That case doesn't exist.

Now you have a potential RPC 3.3 candor issue with the tribunal — a topic covered in a previous post. But here's what most attorneys miss: you may also have an RPC 8.4(c) problem, depending on how that representation was made and how it was relied upon. And if the filing delays proceedings, triggers a motion to strike, or requires opposing counsel to chase phantom citations, you could find yourself explaining to the ACPE why that wasn't conduct prejudicial to the administration of justice under 8.4(d).

The New Jersey Supreme Court's disciplinary arm has not issued a formal opinion specifically addressing AI-generated misconduct as of this writing. But the analytical framework already exists — and it doesn't need a new opinion to apply.

---

Three AI Brief-Writing Habits That Create RPC 8.4 Exposure

1. Treating AI-generated citations as verified until proven wrong. The safer mental model is the reverse: treat every citation as unverified until you've confirmed it yourself in an authoritative legal database. AI tools — even those marketed specifically to lawyers — fabricate citations with remarkable confidence. Your professional obligation doesn't transfer to the vendor.

2. Using AI to characterize opposing counsel's arguments. Some attorneys are now using AI to summarize or respond to the other side's briefs. If the AI mischaracterizes a position — even subtly — and you file that characterization, you've put words in opposing counsel's mouth before a tribunal. That's a misrepresentation problem, even if it was the machine's error.

3. Letting AI write your statement of facts from raw notes. AI is not a neutral drafter of facts. It fills gaps, smooths ambiguities, and sometimes invents coherence where none existed in your source material. A statement of facts that overstates a client's position — because the AI inferred more than your notes supported — is still your statement of facts.

---

What the Practical Safeguard Looks Like

None of this means you shouldn't use AI for brief writing. The efficiency gains are real, and for solo attorneys managing their own calendars, research, and client calls simultaneously, AI-assisted drafting isn't a luxury — it's a survival tool.

But the workflow matters. Before any AI-assisted brief leaves your hands:

• Run every case citation through Westlaw or Lexis. Full stop. Not a spot-check — every one.
• Read every factual assertion against your source documents, not against what sounds right.
• Print the AI's draft and the opposing brief side-by-side if you're using AI to characterize the other side's position. Confirm the characterization yourself.
• Document your review process. If a disciplinary complaint ever comes, you want a contemporaneous record showing deliberate attorney oversight — not just a filing timestamp.

---

The Broader Point

RPC 8.4 is the rule that catches what falls through the cracks of everything else. It's a backstop — which means by the time it applies, something has already gone wrong elsewhere in the chain. For NJ attorneys leaning on AI tools in their litigation practice, the question isn't whether AI can write a good brief. It can. The question is whether the attorney behind it exercised enough genuine judgment that the brief is actually theirs.

That judgment doesn't happen automatically. It has to be built into the workflow, deliberately, before the file gets submitted.