A Closer Look at ABA Formal Opinion 512: What NJ Solo Attorneys Should Actually Do With It

ABA Formal Opinion 512 landed on July 29, 2024, and it immediately got treated as if it were binding law. It isn't. The ABA Committee on Ethics and Professional Responsibility issues formal opinions to interpret the Model Rules, which New Jersey hasn't adopted wholesale. That distinction matters more than most small-firm attorneys realize.

That said, Opinion 512 is the most thorough national statement on generative AI that the legal profession currently has, and a NJ solo who ignores it is leaving real guidance on the table. The trick is reading it carefully: taking what applies, adjusting what doesn't, and knowing the difference.

What Opinion 512 actually says

The opinion addresses generative AI through five Model Rules: 1.1 (competence), 1.6 (confidentiality), 5.1 and 5.3 (supervision), and 5.3's application to AI as a "nonlawyer assistance" question. Its core positions:

• Competence under Model Rule 1.1 requires understanding enough about how a generative AI tool works to use it without inadvertently harming the client. You don't need to be an engineer, but you do need to know what the tool is likely to get wrong.
• Confidentiality under Model Rule 1.6 requires informed analysis before inputting client data into any AI system. The opinion notes that many consumer-grade tools train on user inputs, which can expose client information.
• Supervision duties under Model Rules 5.1 and 5.3 extend to AI-generated work product. A lawyer cannot simply accept AI output as finished work; someone with authority has to review it.

Opinion 512 also explicitly says that whether a particular use of AI requires client disclosure turns on the facts, and that there is no blanket disclosure obligation solely because AI was used.

Where NJ diverges, or adds to this

New Jersey's governing document on this topic is the Supreme Court's "Preliminary Guidelines on New Jersey Lawyers' Use of Artificial Intelligence," issued January 24, 2024. That document predates Opinion 512 by six months and approaches the question differently.

The NJ guidelines are organized around existing RPC obligations rather than Model Rule commentary. They emphasize that all existing professional conduct obligations apply to AI use, without carving out any safe harbors. The guidelines are explicit that competence (NJ RPC 1.1), supervision (NJ RPC 5.3(b)), and confidentiality (NJ RPC 1.6(a)) each independently impose duties that AI use must satisfy.

One meaningful gap: Opinion 512 discusses AI as "nonlawyer assistance" under Model Rule 5.3 at length, treating the AI system itself as something requiring supervisory structure. NJ RPC 5.3(b) places that duty on attorneys who have "direct supervisory authority" over non-lawyer work. The NJ guidelines apply the same framing. So the substantive result is similar, but the path to get there is the state RPCs, not the ABA's Model Rule commentary.

New Jersey has not issued an AI-specific ACPE opinion. The NJ Advisory Committee on Professional Ethics has not addressed generative AI directly as of this writing. Attorneys sometimes cite ACPE Opinion 701 (2006), which covers electronic storage and cloud computing, as background context for cloud-based AI tools. That's fair as analog reasoning. It doesn't substitute for the NJ Supreme Court's guidelines, which are the primary source.

The practical problem Opinion 512 identifies, and NJ hasn't solved

Opinion 512 names something that the NJ guidelines leave implicit: the competence question isn't just about checking AI output for errors. It includes understanding the specific tool's architecture well enough to assess what category of errors it tends to produce.

A general-purpose large language model and a legal-research-specific AI with retrieval-augmented generation fail in fundamentally different ways. A general-purpose model hallucinates citations confidently. A retrieval-augmented tool might retrieve the right case but mischaracterize its holding. The review process that catches one won't reliably catch the other.

NJ RPC 1.1 requires the legal knowledge, skill, thoroughness, and preparation reasonably necessary for the representation. Applying that standard to AI means a NJ attorney needs to know, before deploying a tool, what type of error the tool is prone to, and build a review step that targets that specific failure mode. Opinion 512's framing helps put that obligation in concrete terms, even if it's the NJ RPC, not the ABA opinion, that creates the duty.

What the rule requires

Under NJ RPC 1.1, a lawyer must possess the competence necessary for the representation, which includes understanding tools used in that representation. Under NJ RPC 1.6(a), a lawyer may not reveal information relating to the representation without client consent or a recognized exception. Under NJ RPC 5.3(b), a lawyer with direct supervisory authority over a non-lawyer must make reasonable efforts to ensure that person's conduct is compatible with the lawyer's professional obligations. The NJ Supreme Court's Preliminary Guidelines on New Jersey Lawyers' Use of Artificial Intelligence (January 24, 2024) confirm that these three obligations apply directly to AI use.

What I recommend

Read Opinion 512 as a practical framework, not a rulebook. Its five-rule structure gives you a useful checklist for auditing any AI tool you're considering: Does my use of this tool satisfy competence? Confidentiality? Supervision? Those questions map cleanly onto NJ's own RPCs.

Before you use any AI tool for client work, document in writing which tool you used, why you selected it, what its known failure modes are, and how you reviewed its output. That note doesn't need to be long. It does need to exist, because if something goes wrong, your ability to show a reasonable, informed process is what stands between you and a disciplinary finding.

Don't confuse Opinion 512's statement that there's no blanket disclosure obligation with a conclusion that disclosure is never required in New Jersey. That's a facts-and-circumstances analysis under NJ RPC 1.4. Opinion 512 is persuasive on the framework; the NJ RPC governs the outcome.