An AI policy that is not a ban.

Opinion 512 does not ban AI. It makes the lawyer responsible for the output, across several professional duties: competence, confidentiality, communication, supervision, candor, and fees. The template groups those duties into four practical areas a small firm can actually maintain, with plain commitments under each.

Competence and supervision (Rules 1.1, 5.1, 5.3)

A supervising lawyer reviews AI-assisted work before it leaves the firm, and the review is logged. The reviewer, not the tool, is accountable for the result.

Confidentiality (Rule 1.6)

Client information only enters approved tools with a known data posture. Consumer-grade tools are not a place for client data, and vendor terms are checked before anything goes in.

Communication, drafting, and candor (Rules 1.4, 3.3)

Every AI-generated citation is verified against the primary source before it is filed or sent. Client disclosure is set to your jurisdiction's requirement, not a blanket rule.

Fees and operating discipline (Rule 1.5)

Time saved by AI is not billed as time worked. Efficiency is passed through honestly, and any tool costs stay reasonable and disclosed.

Six pages, ready to adapt. The fields you fill in are marked; everything else is written to be adopted as-is.

• Scope and definitions: what counts as AI, and who the policy binds.
• The core policy: plain commitments under each duty area, with fill-in fields.
• Governance and recordkeeping: an approved-tool register, a supervision log, and a review cadence.
• The documentation 512 rewards: a checklist of artifacts to keep for a client, court, or malpractice carrier.
• An incident-response procedure for when client data reaches an unapproved tool.
• An acknowledgment form for everyone the policy binds.

Solo and small firms that are already using AI and want a policy that is defensible without standing up a committee. Pair it with the free ABA 512 diagnostic to see where your operating model actually stands before you adopt anything.