The NJ Attorney Competence Standard for AI Is Already Here — Most Small Firms Just Haven't Noticed

The ABA made it official back in 2012: competence under Model Rule 1.1 includes keeping up with "the benefits and risks of relevant technology." New Jersey adopted equivalent language, and for years most attorneys interpreted it loosely — know how to use email, maybe e-filing, done.

That era is over.

Generative AI has quietly moved the definition of "technologically competent" into genuinely uncomfortable territory for NJ solo and small-firm attorneys. Not because the Bar has issued a new rule, but because the standard that was always there is now being measured against a dramatically more capable — and more dangerous — class of tools.

---

What NJ RPC 1.1 Actually Requires in an AI Context

New Jersey's RPC 1.1 requires that attorneys provide "competent representation," which includes "the legal knowledge, skill, thoroughness, and preparation reasonably necessary for the representation." The Comment to Rule 1.1 in NJ, tracking the ABA, explicitly pulls technology into that definition.

The word "reasonably" is doing a lot of work here. Reasonableness is a peer-group standard. As AI tools for legal research, contract drafting, and case summarization become more widely adopted by NJ practitioners, the floor of what is "reasonable" rises with adoption. Said plainly: if a meaningful cohort of NJ solo attorneys are using AI to catch issues your manual review misses, your manual-only review starts to look like less than competent preparation.

This isn't speculation. It mirrors exactly what happened with case management software in the early 2000s — tools that were optional novelties became, within a decade, the baseline expectation for organized representation.

---

Three Competence Failures Hiding Inside AI Workflows

The troubling part isn't the attorneys who refuse to use AI. It's the ones who use it carelessly, and thereby create a competence problem they didn't have before.

1. Treating AI output as a finished work product. An NJ solo running a contract review through a general-purpose LLM and delivering those results without independent legal judgment isn't being efficient — they're offloading legal analysis to a tool that has no law license and no liability. RPC 1.1 requires your legal knowledge. AI can surface issues; only you can evaluate them.

2. Using an AI tool without understanding its training cutoff and jurisdiction gaps. Competent use of AI for New Jersey-specific research requires knowing whether the tool has reliable coverage of NJ case law and NJ-specific statutes. A tool trained primarily on federal decisions may miss critical NJ Appellate Division rulings or NJLAD nuances. Asking "did the AI find it?" is not the same as asking "is this actually the law in New Jersey?"

3. No hallucination audit process in place. The post-Mata v. Avianca world has made citation hallucinations a household name among litigators, but transactional attorneys and estate planners face an equivalent trap: AI-generated statutory references, regulatory thresholds, or form language that simply doesn't exist or has been superseded. A competent workflow requires a verification step — specifically checking every cited authority before it leaves your desk.

---

Building a Competence-First AI Protocol (A Practical Starting Point)

You don't need an enterprise AI governance program. A solo or two-attorney firm needs three things:

A documented tool-selection rationale. For each AI tool you use regularly, keep a one-paragraph internal note on why you selected it, what tasks it is approved for, and what it is not used for. This demonstrates deliberate adoption, not casual experimentation.

A task-level review checklist. For each AI-assisted work product — research memo, contract redline, demand letter — define the human verification steps that must occur before delivery. Make it explicit. "Reviewed by attorney" is too vague. "All citations verified against Westlaw/Lexis; NJ-specific provisions confirmed against current NJ statutes" is competence-level documentation.

A continuing education commitment. The NJ Supreme Court's MCLE requirements don't yet mandate AI-specific credits, but the Bar's guidance is trending toward higher expectations. Getting ahead of it now — through CLE courses, Bar publications, or advisory resources — protects you and signals good faith.

---

The Competence Gap Is Also a Business Risk

There's a practical, non-disciplinary reason to take NJ RPC 1.1 seriously in the AI context: malpractice exposure. An AI-generated error that a competent attorney exercising proper oversight would have caught is exactly the fact pattern that makes E&O carriers and plaintiff's malpractice attorneys interested. "I used an AI tool" is not a defense — it may actually be the admission that establishes the breach.

The good news is that the NJ attorneys best positioned in this environment aren't necessarily the most tech-savvy. They're the ones who treat AI adoption the same way they treat any other professional judgment call: with structure, documentation, and a healthy skepticism toward anything they didn't personally verify.

That's not a new skill. It's the oldest one in the profession.