Client Disclosure Templates and NJ's Preliminary AI Guidelines: Is a Generic Consent Form Enough?

The NJ Supreme Court's Preliminary Guidelines on New Jersey Lawyers' Use of Artificial Intelligence (January 24, 2024) tell attorneys to be transparent with clients about AI use in their matters. Most solo practitioners read that sentence, nod, and then paste a one-liner into their retainer agreement. Job done, right?

Probably not. And if a grievance ever lands on your desk, that single-sentence disclosure is going to feel very thin.

What the guidelines actually say about client disclosure

The Preliminary Guidelines don't provide a disclosure template, a minimum word count, or a required timing. What they do say is that attorneys have an obligation to inform clients when AI is being used in a way that is material to the representation. They also tie this obligation back to NJ RPC 1.4(b), which requires a lawyer to explain a matter "to the extent reasonably necessary to permit the client to make informed decisions regarding the representation."

Read those two things together and the picture gets more specific. "Informed decisions" is not satisfied by a buried sentence in a six-page retainer that says "we may use AI tools." That language tells the client nothing about what tools, what tasks, what data is being processed, or what they might want to ask about.

The Preliminary Guidelines are also candid about the fact that AI outputs carry error risks. They specifically note that AI can produce inaccurate information and that a supervising attorney must review and verify AI-generated work product. That risk-of-error context is material. A client deciding whether to approve a strategy built partly on an AI-generated memo would probably want to know that the memo required attorney verification precisely because the underlying tool can and does make things up.

What most small-firm disclosures actually contain

Here's a hypothetical that any NJ solo practitioner can recognize: a retainer agreement has a single paragraph titled "Use of Technology." It says the firm uses technology-assisted tools to improve efficiency and that client data is handled in accordance with applicable law. The attorney considers this adequate AI disclosure.

It isn't, and here's why. That clause doesn't identify the category of tool (generative AI vs. a rule-based document assembler vs. a legal research database), doesn't describe what client information is being input, and doesn't tell the client whether data is retained by a third-party provider. A client who later learns their detailed family law facts were processed through a commercial AI platform that trains on user data would have a legitimate grievance that the consent they gave was not informed consent.

RPC 1.4(b) requires the attorney to explain things to the extent necessary for informed decisions. When AI is doing substantive work, a client has at least three decisions to make: whether to consent to that use, whether to ask about the specific tools, and whether to inquire about what happens to their data. A generic technology clause forecloses all three without ever surfacing them.

Building a disclosure that actually holds up

A defensible AI disclosure doesn't need to be a legal treatise. It does need to answer a few specific questions, and it needs to be delivered at the right time.

On content: identify the category of AI tool being used and the task it's performing. "We may use generative AI tools such as large language models to assist with drafting, research, and document review" is more honest than "technology-assisted tools." Acknowledge that AI outputs are reviewed and verified by the supervising attorney before any reliance. If your AI vendor's terms allow them to use inputs for model training, say so, and tell the client you've taken steps (such as disabling training data settings where available) to limit that exposure.

On timing: a retainer-stage disclosure covers the relationship prospectively, but it can't anticipate every tool. If you add a new AI platform mid-matter, the Preliminary Guidelines' emphasis on ongoing attorney judgment suggests a mid-matter notice is the prudent move, not silence.

On format: a separate, signed AI disclosure addendum is more defensible than a buried retainer clause. It signals that you treated this as a distinct informed-consent conversation, not a boilerplate catch-all.

One practical note on the ACPE: as of this writing, the Advisory Committee on Professional Ethics has issued no AI-specific formal opinion. ACPE Opinion 701 (2006) remains the governing guidance on electronic storage of client information and cloud-based tools, but it predates generative AI entirely. That gap means the Preliminary Guidelines and the underlying RPC text are currently your primary reference points, not a neatly packaged ACPE advisory.

What the rule requires

NJ RPC 1.4(b) requires a lawyer to explain a matter to the extent reasonably necessary for the client to make informed decisions. The NJ Supreme Court's Preliminary Guidelines on New Jersey Lawyers' Use of Artificial Intelligence (January 24, 2024) require attorneys to be transparent with clients about AI use and to review and verify AI-generated work product before relying on it.

What I recommend

Retire the generic technology clause and replace it with a standalone AI disclosure addendum, signed separately from the retainer. Identify the category of tool, the tasks it performs, the attorney's verification responsibility, and any data-retention or training implications in your vendor's terms. Deliver a supplemental notice any time you add a materially different AI tool mid-matter. Keep a copy of both the disclosure and the client's acknowledgment in the matter file. None of this is burdensome, and all of it is far easier to defend than silence.