What Happens When NJ's Preliminary AI Guidelines Meet a Misconduct Investigation? A Walkthrough for Solo Attorneys

The NJ Supreme Court issued its "Preliminary Guidelines on New Jersey Lawyers' Use of Artificial Intelligence" on January 24, 2024. Most commentary since then has focused on what the guidelines say attorneys should do. Far less attention has gone to a harder question: what do those guidelines actually mean for a solo attorney who ends up the subject of a disciplinary complaint after something goes wrong with an AI-assisted matter?

That's the walkthrough worth doing now, before the scenario is real.

The guidelines are not a disciplinary rule

This matters more than it might seem. The Preliminary Guidelines are issued by the Supreme Court as guidance, not as a formal amendment to the New Jersey Rules of Professional Conduct. The RPCs themselves, found at Rule 1:14 of the Court Rules, are the operative disciplinary framework. When the Office of Attorney Ethics receives a grievance, investigators look to the RPCs, not to guidance documents, as the basis for a charge.

What that means practically: an attorney cannot defend against a misconduct finding by pointing to the guidelines and saying "I followed those." Nor can the OAE charge a violation of the guidelines as a standalone offense. The guidelines function more like an interpretive lens, one that shapes how the existing RPCs get read in the context of AI use.

So if a client files a grievance because an AI-generated brief contained a hallucinated statute that you filed without catching, the charge isn't "violation of the AI guidelines." It's a potential violation of NJ RPC 3.3(a)(1) (false statement of law to a tribunal) or NJ RPC 1.1 (competence), or both. The guidelines just inform what "competence" looks like in 2024 and beyond.

What the guidelines actually flag for investigators

The Preliminary Guidelines do several things that become relevant in a disciplinary context. They identify categories of risk the Supreme Court considers material: accuracy verification, confidentiality of client information, supervision of AI outputs, and transparency with courts. When the OAE evaluates a complaint involving AI use, those categories are exactly the fault lines they'll examine.

Specifically, the guidelines state that attorneys remain responsible for the accuracy of any AI-generated work product. That language mirrors the responsibility principle already embedded in NJ RPC 5.3(b), which requires attorneys to take reasonable steps to ensure that the conduct of non-lawyer assistance is compatible with the attorney's professional obligations. Courts have analogized AI tools to non-lawyer assistance in this context, and the guidelines reinforce that analogy without resolving all of its edges.

What the guidelines don't do is quantify "reasonable." They don't tell you how many times you need to verify a citation, what a compliant review workflow looks like, or whether using a particular AI tool's hallucination rate makes your verification process adequate or inadequate. That gap matters enormously in a disciplinary context because the OAE evaluates conduct against a reasonableness standard, and right now there's no bright line to point to.

The investigative process itself

For solo attorneys who haven't been through an OAE investigation, the procedural basics are worth knowing. Under Rule 1:20-3, the OAE investigates grievances filed by clients, opposing counsel, courts, or any member of the public. The attorney receives written notice and has the right to respond. If the OAE finds probable cause, the matter proceeds to either a district ethics committee hearing or, in serious cases, directly to the Disciplinary Review Board.

Here's where AI use creates a documentation problem that most solos haven't thought through. If the OAE asks you to explain your review process for an AI-generated document, what can you actually show them? A solo attorney who used an AI tool, accepted its output, and filed without a documented verification step will have difficulty demonstrating that their process was reasonable, even if they subjectively believed it was. The guidelines' emphasis on attorney responsibility for AI outputs creates an implicit expectation that some verifiable review happened.

A log of your review steps, even something as simple as a dated file note or a checklist in your case management system, is worth more in that moment than any after-the-fact explanation.

One procedural protection that hasn't changed

Nothing in the Preliminary Guidelines alters the attorney's fundamental due process rights in a disciplinary proceeding. The burden of proof in NJ disciplinary matters remains on the presenter, and clear and convincing evidence is required for most sanctions under the standard established by the Disciplinary Review Board's longstanding practice.

That standard is meaningful. An isolated AI-related error, promptly corrected and accompanied by evidence of a good-faith review process, is a very different disciplinary profile than a pattern of filing unverified AI outputs across multiple matters. The guidelines don't change that calculus, but they do set an expectation against which your conduct will be measured going forward.

What the rule requires

NJ RPC 3.3(a)(1) prohibits making a false statement of material fact or law to a tribunal. NJ RPC 1.1 requires competent representation, which includes the legal knowledge, skill, thoroughness, and preparation reasonably necessary for the representation. NJ RPC 5.3(b) requires a lawyer having direct supervisory authority over non-lawyer assistance to make reasonable efforts to ensure the person's conduct is compatible with the professional obligations of the lawyer. The NJ Supreme Court's Preliminary Guidelines on New Jersey Lawyers' Use of Artificial Intelligence (January 24, 2024) state that attorneys remain responsible for the accuracy of AI-generated work product and must verify outputs before use.

What I recommend

Build a paper trail now, while it's routine rather than reactive. For every matter where you use an AI tool to generate or substantially assist with a court filing, keep a brief written note in the file documenting what you verified and how. It doesn't need to be elaborate: the tool used, the specific output reviewed, the verification step taken (cross-checked citations against Westlaw, confirmed statutory text on the official code site, etc.), and the date.

If an OAE investigator ever asks what your review process looked like, that contemporaneous note is your answer. The absence of one leaves you reconstructing your reasoning from memory, which is a much weaker position, especially as the guidelines sharpen the expectation that AI outputs get checked before they leave your office.