What NJ Attorneys Get Wrong About RPC 1.1 Competence When They Choose an AI Tool

Most New Jersey attorneys think about RPC 1.1 in terms of subject matter: you need to know family law to handle a divorce, real estate law to close a transaction. That framing is correct but incomplete. Comment 8 to ABA Model Rule 1.1, which New Jersey's version tracks in spirit, makes clear that competence includes "keeping abreast of changes in the law and its practice, including the benefits and risks associated with relevant technology."

That one phrase is doing a lot of work right now, and most solo attorneys in NJ aren't reckoning with it seriously when they select an AI tool.

The Selection Decision Is Already a Competence Decision

Picking an AI writing or research tool isn't a vendor procurement task you can treat like choosing a coffee maker for the office. The moment you rely on an AI output to draft a brief, analyze a contract, or summarize a deposition, that tool is embedded in your professional judgment. And under RPC 1.1, your professional judgment is always on the hook.

The problem is that most solo attorneys make the selection decision based on two criteria: price and familiarity. They choose the tool a colleague mentioned at a bar association event, or they try the one that has a legal-sounding name. Neither of those proxies tells you anything useful about whether the tool is appropriate for the specific task you're using it for.

Here's the distinction that matters. A general-purpose large language model like ChatGPT or Claude is optimized for fluency, not legal accuracy. A purpose-built legal AI tool like Westlaw Precision or CoCounsel is trained on curated legal data and includes citation-checking infrastructure. Using the former to research whether a specific NJ statute preempts a local ordinance, without any independent verification workflow, is a different competence risk than using the latter for the same task. They're not interchangeable, and your RPC 1.1 obligation requires you to understand that difference before you pick one.

Three Questions That Actually Map to RPC 1.1

Before committing to any AI tool for client-facing work, a NJ solo practitioner should be able to answer these questions:

What data did this tool train on, and when was that data cut off?

New Jersey law changes. Court rules get amended. The Appellate Division issues opinions that shift how trial courts apply standards. An AI tool with a training cutoff of 18 months ago has no knowledge of any of that. If you're using it to analyze current NJ civil procedure questions, you're working with stale information and you may not know it. Competence requires knowing the limitation.

Does this tool cite sources I can independently verify?

A tool that generates a confident paragraph about NJ contract doctrine without a citation is not a research tool; it's a drafting assistant. Those are different things with different risk profiles. If you're treating the former as the latter, you've made a competence error before you've typed a single word into a client document.

What happens to the text I submit as a prompt?

This question bridges RPC 1.1 and RPC 1.6, but it lives in competence first: you can't competently use a tool you don't understand. If the tool you're using sends your prompts to a third-party model for processing, stores them for product improvement, or doesn't offer a business associate agreement for health-adjacent matters, you need to know that before you paste in anything client-specific. Not knowing isn't neutral; it's a competence gap.

The NJ-Specific Context Worth Knowing

The New Jersey Supreme Court's Committee on the Courts of the Future has been actively examining AI's role in legal practice, and the NJ State Bar Association has signaled increased attention to technology-related ethics questions at the practitioner level. While NJ has not yet issued a standalone formal ethics opinion on AI tool use (unlike some other state bars), the existing RPC framework, particularly RPC 1.1 read alongside RPC 5.3 on non-lawyer supervision, already gives the ethics infrastructure everything it needs to evaluate how an attorney selected and used an AI tool.

That means the documentation you keep (or don't keep) about why you chose a tool, how you tested it before using it on client matters, and how you verified its outputs is already relevant to your ethical posture today, not at some future date when formal guidance arrives.

A Simple Pre-Adoption Checklist

Before using any AI tool for the first time on a client matter, answer these five questions in writing (even just a note in your file):

1. What specific task is this tool being used for?
2. What is this tool's training data source and knowledge cutoff date?
3. How will I independently verify any legal citations or conclusions it produces?
4. What are this vendor's data handling and confidentiality terms?
5. Do I have enough understanding of how this tool works to explain its output to a client or a court if asked?

If you can't answer all five, you're not ready to use the tool on client work. That's not a high bar; it's the floor that RPC 1.1 has always set. The technology is new, but the obligation isn't.

The attorneys who will stay on the right side of this rule aren't the ones who avoid AI entirely. They're the ones who treat tool selection as a professional decision requiring deliberate judgment, the same way they'd treat choosing an expert witness or deciding which legal research database to rely on for a critical motion.