NJ Ethics Opinion 742 Has a Blind Spot — And It Could Cost You a Client

New Jersey lawyers are generally aware that the Advisory Committee on Professional Ethics (ACPE) issued Opinion 742 addressing cloud computing and third-party data storage. Many practitioners treat it as their governing document for AI tool adoption. That instinct is understandable — but dangerously incomplete.

Opinion 742 was written for a world of Dropbox and Google Drive. It was not written for a world of large language models that ingest your prompts, potentially train on your data, and return outputs that may be confidently, fluently wrong. Using it as your AI compliance framework is like using a 2009 GPS to drive a 2025 road.

Here's where the real exposure lives.

---

The Competence Problem Is More Specific Than You Think

RPC 1.1 requires competent representation, and Comment 8 (adopted in New Jersey) explicitly calls for lawyers to keep pace with "changes in the law and its practice, including the benefits and risks associated with relevant technology."

Most NJ practitioners read this as: "I should know what AI is." The actual standard is harder. Competence means you must understand:

• How a specific tool generates its output — is it retrieval-based, generative, or a hybrid?
• Whether the tool has a knowledge cutoff that may miss recent NJ Appellate or Supreme Court decisions
• Whether citations it produces have been verified against a primary source — not just hallucination-checked by another AI layer

A solo attorney in a South Jersey plaintiff's practice recently shared with me that their AI-assisted brief contained a citation to a New Jersey Appellate Division case that was reversed on appeal six months prior. The AI had no awareness of the subsequent history. The error made it past two review passes before a sharp paralegal caught it. That near-miss is not an edge case. It is a routine risk for any practitioner who does not have an explicit citation-verification step baked into their workflow.

---

RPC 1.6 and the Vendor Agreement Nobody Reads

New Jersey's RPC 1.6 requires reasonable efforts to prevent unauthorized disclosure of client information. The ACPE, consistent with Opinion 742, applies a "reasonable measures" standard — not a perfection standard. But "reasonable" is doing a lot of work here, and courts are beginning to define its edges.

When a New Jersey solo or small firm subscribes to an AI legal research or drafting tool, there are typically three documents that govern what happens to client data: the Terms of Service, the Data Processing Agreement (DPA), and the Privacy Policy. In my consulting work, I have reviewed dozens of these. The variations are striking:

• Some enterprise-tier tools contractually commit to zero data retention for prompts and outputs.
• Some standard-tier tools (often the ones with the lowest price point) reserve the right to use your inputs for model improvement.
• Some tools are ambiguous enough that the answer requires a direct inquiry to the vendor's legal team.

If you are using the free or prosumer tier of any AI tool in client matters, you likely have not executed a DPA. That is a RPC 1.6 problem — not a theoretical one, but an operational one you can fix this week.

---

A Practical Three-Step Posture for NJ Practitioners

You do not need a 40-page AI policy to operate responsibly. You need a coherent posture. Here is a minimal viable framework:

1. Tier your tools. Designate which AI tools are approved for client-matter use and which are internal-only. This is a five-minute conversation that most small firms have never had explicitly.

2. Execute or obtain a DPA before inputting any client-identifiable information. If the vendor does not offer one, that tier of the product is not appropriate for client work. Full stop.

3. Build a hard verification gate into every AI-assisted work product. Any case citation, statutory reference, or regulatory cite produced by an AI must be independently verified in Westlaw, Lexis, or directly on the NJ Courts website — by a human — before filing or sending. This step belongs in your written workflow, not just your memory.

---

The Bar Is Watching, and the Standard Is Moving

The New Jersey Supreme Court has signaled through its Committee on the Courts of the Future that AI guidance will continue to evolve. Other jurisdictions — including Florida, California, and New York — have issued more granular AI-specific opinions. NJ's more detailed guidance is coming.

Practitioners who build sound data governance habits now will not have to scramble to retrofit compliance when that guidance lands. More importantly, their clients will not bear the cost of the scramble.

The competitive advantage in the next three years does not belong to the firm that adopts AI fastest. It belongs to the firm that adopts it most responsibly — and can prove it.

---

Adam Elias is the founder of Elias Advisory LLC, helping solo lawyers and small law firms adopt AI securely and operationally. Questions or topics for a future Sidebar post? Get in touch.