Does AI-Assisted Conflicts Screening Actually Hold Up Under NJ Ethics Rules?

Conflicts screening has never been glamorous. For solo attorneys and small NJ firms, it's often a manual crawl through a spreadsheet, an old intake form, or — if you're being honest — memory. AI vendors are now pitching automated conflicts-check features as a core selling point. Some of them are genuinely useful. Some of them will give you false confidence that could end your career.

Before you automate anything, you need to understand exactly what NJ ethics rules demand — and where the gap between "the software flagged it" and "you actually checked" becomes your problem.

What NJ RPCs 1.7, 1.9, and 1.10 Actually Require

New Jersey's conflicts framework isn't complicated in theory, but it's unforgiving in practice.

RPC 1.7 prohibits concurrent conflicts — representing a client whose interests are materially adverse to another current client, or where your own interests could impair your representation. RPC 1.9 extends that duty to former clients: you can't represent someone in a matter that is the same or substantially related to a prior matter if the new client's interests are materially adverse to the former client. RPC 1.10 then imputes those conflicts across the entire firm.

The standard these rules demand isn't "I ran it through the software." It's reasonable diligence — a term the NJ Supreme Court's Advisory Committee on Professional Ethics has interpreted to require actual judgment about the relationships involved, not just name-matching. A conflicts check is not a lookup. It is an analysis.

Where AI Screening Tools Genuinely Help

That said, dismissing AI from your conflicts workflow would be a mistake in the other direction. Here's where it adds real value:

Database coverage and speed. A well-configured AI-assisted system connected to your practice management software can surface name variants, related entities, and affiliated parties far faster than manual review. If a new client's LLC shares a registered agent with an adverse party from two years ago, a good system catches that. Your memory probably doesn't.

Adverse party identification. Some legal AI platforms (Clio, MyCase, and purpose-built tools like Confido Legal) now pull adverse party names directly from intake forms and prior matter data, flagging potential RPC 1.9 issues before the first consultation call. This is legitimately useful for solo practitioners juggling 40+ active matters.

Documentation. AI tools create a timestamped record of when the check was run, against what data, and what it returned. In a disciplinary proceeding, that paper trail matters. "I checked" is far less persuasive than a log showing you checked on a specific date against a specific dataset.

Where AI Screening Tools Fall Dangerously Short

Here is where NJ solo attorneys need to slow down — and where vendor marketing gets ahead of legal reality.

Name-matching is not conflict analysis. AI systems identify potential overlaps. They do not evaluate whether a matter is "substantially related" under RPC 1.9, whether a client's interests are "materially adverse," or whether a screen is sufficient under RPC 1.10(c). Those are legal judgments. An algorithm can hand you a flag; it cannot hand you a conclusion.

Your data is only as good as your intake. If your prior matter records are incomplete — adverse parties were never entered, corporate affiliates were never listed, co-defendants were never linked — the AI will return a clean result on a dirty dataset. Garbage in, malpractice out. Before you trust any automated system, audit your historical data first.

Waiver and consent still require a lawyer. Even when a conflict is identified and is waivable under RPC 1.7(b), obtaining informed consent requires a written explanation of the conflict and the risks involved. No AI tool does this for you. The confirmation letter has to come from you, reviewed by you, tailored to the specific facts.

A Practical Workflow for NJ Small Firms

Here's a defensible approach that integrates AI without delegating the legal judgment:

1. Run the automated check first — use your practice management system's built-in conflicts tool or a dedicated intake automation to flag names, entities, and related parties at the moment of intake.
2. Review every flag personally — don't let a paralegal or intake coordinator clear a conflict flag. Under RPC 5.3, you are responsible for their work product, but a conflicts determination is a legal conclusion.
3. Document the analysis, not just the result — your file note should say why a flagged item was cleared, not just that it was cleared. One sentence of reasoning is enough. Zero sentences is not.
4. Run a second pass before filing anything — circumstances change between intake and the first court appearance. New adverse parties surface. Run the check again.
5. Audit your historical data at least annually — set a calendar reminder. Clean data is what makes AI screening reliable.

The Bottom Line

AI-assisted conflicts screening is a tool for thoroughness, not a substitute for analysis. NJ RPCs 1.7, 1.9, and 1.10 place the duty squarely on the attorney — not the software. The firms that will use these tools well are the ones that understand precisely where the algorithm ends and the professional judgment begins. The ones that get into trouble are the ones who click "no conflicts found" and move on without reading why.

Use the machine. But own the conclusion.