Comparing the NJ Supreme Court's AI Guidelines to ACPE Opinion 701: Where the Two Actually Overlap
Photo by Ambrose Prince on Unsplash
6 minOctober 31, 2026

Comparing the NJ Supreme Court's AI Guidelines to ACPE Opinion 701: Where the Two Actually Overlap

NJ Supreme Court AI GuidelinesACPE Opinion 701NJ law firm data security

Drafted by AI, reviewed by Adam Elias. Commentary, not legal advice. Published October 31, 2026.

Most NJ practitioners 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 documents from two completely different worlds. One is about AI. One is about putting files in the cloud back when Dropbox was a startup. Close the opinion, file it away, move on.

That reading is too convenient.

When a solo attorney uses a cloud-hosted AI tool to draft a motion, summarize a deposition, or screen a contract, both documents are live at the same time. Understanding where they reinforce each other, and where they create separate requirements, is now a basic operational question for any NJ small firm.

What the documents actually say, side by side

ACPE Opinion 701 addressed electronic storage of client files, including storage hosted by third-party vendors. Its core holding: an attorney may store client information with a third-party electronic storage vendor provided the attorney exercises reasonable care to ensure the vendor maintains adequate security measures. The opinion identifies the duty as ongoing, not a one-time vendor-selection checkbox. It also requires the attorney to stay alert to breaches, to notify clients of any breach affecting their information, and to retain the ability to retrieve files if the vendor relationship ends.

The Preliminary AI Guidelines, issued nearly two decades later, sit on top of that baseline. They instruct NJ attorneys that existing RPCs apply fully to AI use, and they explicitly identify RPC 1.6 confidentiality duties as operative. They state that attorneys must understand the AI tools they use, must take steps to prevent inadvertent disclosure of client information, and must evaluate whether a given AI tool's data practices are consistent with those duties.

The overlap is direct: both documents focus on what happens to client data when it leaves your office and sits on someone else's infrastructure. ACPE Opinion 701 set the floor in 2006. The Preliminary Guidelines confirmed in 2024 that the floor is still there and that AI tools don't get a special exemption from it.

The specific gap most attorneys miss

ACPE Opinion 701 requires ongoing attention to security. The Preliminary Guidelines require attorneys to understand how their AI tools handle data, including whether prompts and uploaded documents are used to train the vendor's models.

These two requirements, read together, create a monitoring obligation that is more demanding than either document states alone. It is not enough to review a vendor's terms of service once at signup. If a vendor changes its data retention or training policies (and several major AI vendors have updated those policies multiple times since 2023), an attorney relying on that tool has a continuing duty, grounded in Opinion 701's ongoing-care standard, to catch those changes.

A hypothetical that illustrates the gap: suppose a solo family law attorney subscribes to a cloud AI drafting tool in January, reviews the privacy policy, and concludes that client data is not used for model training. In August, the vendor quietly updates its terms to allow use of user inputs for "product improvement." The attorney never sees the update. Under Opinion 701's ongoing-care standard, that attorney may have a problem, because the duty of reasonable care didn't expire at the moment of initial review.

The Preliminary Guidelines don't spell out a monitoring schedule. Opinion 701 doesn't address AI at all. Neither document, standing alone, tells you exactly what to do. Together, they strongly suggest that a one-time vendor review is insufficient for cloud AI tools whose terms are subject to change.

A note on what the ACPE has and hasn't said about AI

The Advisory Committee on Professional Ethics has not issued an AI-specific opinion as of the date of this post. The Preliminary AI Guidelines came from the NJ Supreme Court directly, not from the ACPE. This matters because some attorneys are waiting for the ACPE to weigh in with a formal opinion before adjusting their practices. That wait may be longer than expected, and the existing combination of Opinion 701 and the Preliminary Guidelines already creates actionable duties.

What the rule requires

Under ACPE Opinion 701, a NJ attorney using a third-party electronic storage vendor (including cloud-hosted AI platforms) must exercise reasonable ongoing care to ensure the vendor maintains adequate security, must be able to retrieve client files if the relationship ends, and must notify clients of any breach. Under the NJ Supreme Court's Preliminary AI Guidelines, existing RPCs, including the confidentiality obligations in RPC 1.6, apply to AI use, and attorneys must take reasonable steps to understand how their AI tools handle client data.

What I recommend

First, treat your AI vendor review the same way you treat a recurring docket reminder: set a calendar item every six months to re-read the vendor's data use and retention policies. Vendors update their terms without fanfare, and Opinion 701's ongoing-care standard doesn't give you a pass because the change was buried in an email you didn't open.

Second, build a one-page internal log for each AI tool your firm uses. Record the date you reviewed the privacy policy, the key terms on data retention and training opt-outs, and the date of your next scheduled review. This costs you an hour to set up and creates a contemporaneous record that demonstrates the kind of reasonable ongoing care Opinion 701 describes.

Third, if your AI tool offers a setting to opt out of data use for model training, turn it on and document that you did so. Not every tool offers this, and the setting alone isn't the entire analysis, but it is the most concrete step you can take right now to align with both the Opinion 701 security-care standard and the Preliminary Guidelines' directive to prevent inadvertent disclosure.

The Preliminary Guidelines are preliminary in name. The duties underneath them, grounded in Opinion 701 and RPC 1.6, are not.

Sources

Read the source before you rely on anything above. The section headed "What the rule requires" states only what these sources require; "What I recommend" is Adam's advice, not the rule.

Get the weekly roundup

New AI Sidebar articles delivered to your inbox. No spam, unsubscribe anytime.