home  /  library  /  aba-formal-opinion-512
Ethics opinions & regulator guidance

ABA Formal Opinion 512: Generative Artificial Intelligence Tools

ABA Standing Committee on Ethics and Professional Responsibility · 2024 · American Bar Association

GovernanceVerificationBilling & feesConfidentialityEvery lawyer, and mandatory reading before writing a firm policy
Why it is on the shelf. Not binding anywhere on its own, and the document that shaped the landscape. It established that no new rules were needed because the existing ones already reach this — a framing that has largely held.

The Institute's reading

Formal Opinion 512 did something more useful than announcing new rules: it declined to. Its central move is that generative AI raises no novel category of ethical obligation, because competence, confidentiality, communication, candour to tribunals, supervision and reasonable fees already reach every question a firm is likely to have. That framing has proved durable, and most state guidance since has followed it rather than departed from it.

Four threads matter most in practice. Competence now includes a reasonable understanding of the benefits and risks of tools a lawyer uses — not technical expertise, but enough to know how they fail. Confidentiality requires evaluating whether inputting client information into a tool constitutes disclosure, which turns on the specific terms governing that tool rather than a general answer. Communication may require telling clients about AI use, depending on the engagement and the materiality. And on fees, the opinion is unusually direct: a lawyer may not bill for time not spent, so efficiency gained through AI cannot be billed as though the hours were worked, and the cost of the tool itself may only be passed to a client under the ordinary rules on expenses.

The limit to note is jurisdictional. This is an ABA opinion, which binds nobody by itself; what governs a lawyer is their own jurisdiction's rules and guidance. A number of states have issued opinions that follow 512 closely and several depart from it in specific respects. Read it as the map, then read your own state.

Key propositions

  • Existing rules of professional conduct already address generative AI; no new category of obligation was created.
  • Competence includes a reasonable understanding of the benefits and risks of the tools a lawyer uses.
  • Whether inputting client information is a disclosure depends on the specific terms governing the tool.
  • A lawyer may not bill hours not worked, so AI-driven efficiency cannot be billed as time spent.
  • The cost of AI tools may be charged to clients only under the ordinary rules governing expenses.
  • Supervisory duties extend to AI use by lawyers and non-lawyer staff alike.

In practice

  • A firm AI policy should be organised around the existing conduct rules rather than as a separate technology document.
  • Billing practice needs an explicit decision, because the opinion forecloses the comfortable one.
  • Vendor terms become an ethics question, not just a procurement question.

Where authorities disagree

State opinions diverge on client disclosure in particular — some read the communication duty as requiring notice of material AI use, others treat it as ordinarily unnecessary where the lawyer remains responsible for the work. There is also live disagreement about how much a lawyer must understand: 512 sets a reasonable-understanding standard, and commentators differ sharply on whether that is satisfied by knowing that a tool can hallucinate or requires something closer to understanding why. The practical answer is that the more your firm relies on a tool, the more the standard asks of you.

AI adoption conciergeorientation · not legal or ethics advice
Happy to dig into it. What would you like to pressure-test from ABA Formal Opinion 512: Generative Artificial Intelligence Tools: one of its propositions, how it applies to your situation, or where it disagrees with the rest of the shelf?