Can the NJ Supreme Court's Preliminary AI Guidelines Coexist With ACPE Opinion 701? A Practical Conflict Map for Solo Attorneys

Most NJ solo attorneys treat the Supreme Court's "Preliminary Guidelines on New Jersey Lawyers' Use of Artificial Intelligence" (January 24, 2024) and ACPE Opinion 701 (2006) as two items on the same compliance checklist. Read one, check the box. Read the other, check the box. Done.

That's not the right way to read them together, and the gaps between them carry real professional risk.

The two sources were written by different bodies, under different mandates, almost two decades apart. ACPE Opinion 701 was issued by the Advisory Committee on Professional Ethics in 2006, interpreting what NJ RPC 1.6 required when attorneys stored confidential client information electronically, specifically in cloud-based or third-party server environments. The Preliminary AI Guidelines were issued by the Supreme Court itself in January 2024, covering a broader set of concerns: competence, confidentiality, supervision, candor, and fees, as they apply to AI tools specifically.

The overlap looks clean on the surface. Both touch confidentiality. Both invoke a "reasonable efforts" standard for protecting client data. But the moment you get specific about what each authority actually says, the alignment starts to show its limits.

Where Opinion 701 still controls the cloud question

ACPE Opinion 701 established that NJ RPC 1.6 permits storing client files on third-party servers, provided the attorney takes "reasonable precautions" to protect confidential information. The opinion lists factors attorneys should weigh: the provider's terms of service, encryption practices, access controls, and what happens to data if the relationship ends. Those aren't suggestions; they're the framework Opinion 701 says attorneys must work through before trusting a third-party system with client files.

Here's the tension: most AI tools that NJ solo attorneys are actually using (think document drafters, contract reviewers, legal research assistants) don't just process data locally. They send information to cloud-based large language models. That means every document you feed into one of those tools may also implicate Opinion 701's checklist, not just the AI Guidelines.

The Preliminary AI Guidelines acknowledge confidentiality but do not supersede Opinion 701. They mention that attorneys must protect confidential information when using AI, but the Court did not issue the Guidelines as a formal ethics opinion and explicitly called them preliminary. The ACPE's cloud-storage framework from Opinion 701 remains independently operative.

Where the two sources actually diverge

The Preliminary AI Guidelines introduce a concept that Opinion 701 doesn't touch: ongoing oversight. The Court's Guidelines describe AI use as requiring "ongoing" attention to accuracy and reliability. That temporal dimension, the idea that your duty doesn't end when you hit send, has no direct parallel in Opinion 701, which focuses on the threshold decision to use a third-party service rather than the continuous use of it.

That gap matters practically. A solo attorney who set up a cloud file storage system in 2015 and ran through the Opinion 701 checklist at that time probably thought the analysis was done. The Preliminary AI Guidelines say, in effect, that the analysis for AI tools is never done. The obligation to monitor, verify, and assess runs throughout the representation.

The two sources also diverge on supervision. Opinion 701 is silent on staff use of cloud tools; it addresses the attorney's own decision to store files. The Preliminary AI Guidelines explicitly extend to "lawyers and their employees," making clear that an attorney is responsible when a paralegal or legal assistant uses an AI tool, not just when the attorney uses it personally.

A hypothetical worth working through

Imagine a solo family law attorney in Trenton who subscribes to an AI drafting tool. The tool is cloud-hosted, processes documents on remote servers, and retains inputs for model improvement unless the attorney opts out. Under Opinion 701, that attorney should have reviewed the provider's data retention terms, confirmed encryption standards, and evaluated what happens to client data if the subscription lapses. Under the Preliminary AI Guidelines, the attorney also needs to understand how the tool generates output, assess its accuracy in the specific context of family law drafting, and supervise any staff who use it.

Neither source alone covers both sets of concerns. That's the conflict map in one paragraph.

A note on what the ACPE has not done

The ACPE has not issued an AI-specific ethics opinion as of the date of this post. That means Opinion 701 remains the most directly applicable ACPE guidance for the data-handling side of AI use. The absence of a targeted AI opinion from the ACPE is itself significant: solo attorneys cannot assume that the Preliminary AI Guidelines fill every gap the ACPE would eventually address. The two authorities may yet diverge further once the ACPE speaks to AI directly.

---

What the rule requires

NJ RPC 1.6, as interpreted by ACPE Opinion 701, requires attorneys to take reasonable precautions before storing client confidential information with a third-party provider, including reviewing the provider's terms, encryption, access controls, and data-recovery procedures. The NJ Supreme Court's Preliminary AI Guidelines on Lawyers' Use of Artificial Intelligence (January 24, 2024) require attorneys to understand the AI tools they use, protect client confidential information when using them, exercise ongoing oversight of AI-generated work product, and extend supervision to employees who use AI tools. The Guidelines are explicitly preliminary and do not constitute a formal ethics opinion.

What I recommend

Before using any cloud-hosted AI tool on client matters, run Opinion 701's checklist and the AI Guidelines' competence and confidentiality requirements as two separate analyses, not one. Specifically: confirm the vendor's data retention settings and opt out of training-data use if the option exists; document that you reviewed those terms; and set a calendar reminder to re-check them annually, because vendor terms change and the Guidelines' "ongoing" language gives that review real teeth. If you have a paralegal or assistant using the tool, your supervision obligation under the Guidelines applies from day one, not after a problem surfaces.