Conflicts Screening With AI at a NJ Small Firm: What RPC 1.7 Actually Requires You to Verify

Conflicts screening is one of those tasks that feels administrative until it blows up. Then it's an ethics problem, a malpractice exposure, and sometimes a fee forfeiture wrapped into one. AI tools have made it genuinely easier to search intake records, match party names, and flag potential overlaps. But "easier" and "sufficient" are not the same thing, and RPC 1.7 does not grade on a curve because you ran a query.

Here's what's actually happening at a lot of NJ small firms right now: an attorney inputs a new client name into an AI-assisted practice management tool, the tool surfaces no conflicts, and intake moves forward. The attorney treats the AI output as the answer. It isn't. It's a starting point.

What RPC 1.7 Actually Asks

New Jersey RPC 1.7 prohibits representing a client if the representation involves a concurrent conflict of interest, which means either a direct conflict between current clients or a significant risk that the representation will be materially limited by the attorney's responsibilities to another client, a former client, or a third party. The rule requires the attorney to make a competent, individualized judgment. That judgment cannot be fully delegated to software.

The competence requirement under RPC 1.1 reinforces this. An attorney using an AI conflicts tool without understanding its limitations isn't meeting the standard of care. Competence includes knowing what your tools can and can't do.

Where AI Conflicts Screening Actually Falls Short

The failure modes are specific, and knowing them helps you build around them.

Name matching is fragile. AI tools search what's in your system. If a related entity, a parent company, an alter ego, or a principal appears under a slightly different name or wasn't entered consistently at intake, the tool won't flag it. A client you represent in a real estate closing might have a corporate affiliate that's adverse to another current client in a contract dispute. If the affiliate name wasn't captured at intake, no AI search finds it.

Relational conflicts don't surface automatically. RPC 1.7 cares about more than direct party-to-party conflicts. Significant personal relationships, financial interests, referral dependencies, and co-counsel arrangements can all create material limitation conflicts. None of those are reliably stored in a practice management database.

Lateral hire conflicts require separate protocol. When an attorney joins your firm, their prior client list may exist only in their memory or a prior firm's system. AI can only screen what it has access to. Conflicts from a lateral's prior practice don't appear in your system unless you build a structured intake process to capture and screen them.

Real-time adverse party updates don't happen. A client you represented two years ago in an unrelated matter may now be the opposing party in something you're taking on today. AI screening against historical records helps here, but only if those records are complete, accurately categorized, and searchable in the way your query assumes.

A Practical Workflow That Actually Holds Up

The goal isn't to abandon AI screening. It's to treat it as the first pass rather than the final word.

Start with your AI tool and document the search parameters you used: what names, what entities, what date range. This creates a record showing you ran a systematic check.

Then require human review of every new matter before intake is confirmed. That review should include a verbal or written question to the incoming client: identify all parties, all related entities, and all individuals with a material interest in the matter. Don't rely on the client intake form alone. People omit things.

For corporate or business matters, run a quick UCC or entity search to identify principals and affiliated entities, then screen those names separately. This takes fifteen minutes and catches a category of conflicts that AI tools routinely miss.

Build a laterals protocol now, before you need it. Any attorney joining your firm should complete a written disclosure of prior clients and matters. Screen that list manually against your current client database the day they start, not after their first client meeting.

Finally, document the entire process. If a conflict is later alleged, your defense is the documented workflow: what you searched, what you asked, what you found, and what judgment you made. An undocumented "I ran a quick check" is very hard to defend.

The Accountability Gap

AI vendors selling conflicts screening features don't bear professional responsibility under the RPCs. You do. The NJSBA and the Office of Attorney Ethics haven't issued a formal opinion specifically on AI-assisted conflicts screening as of this writing, but the existing framework is clear enough: the tool assists the attorney, the attorney answers for the result.

The practical takeaway is narrow and specific. Before you close your next intake, ask yourself whether your AI conflicts check captured every entity name, every related party, and every lateral history that could matter. If the answer is anything other than a confident yes, keep looking before you sign the engagement letter.