Legal information, not legal advice · Every entry verified to its primary source · Independent of any bar association

Michigan AI Ethics Rules for Lawyers

Last updated July 1, 2026 · First published July 1, 2026 · By MHSB Solutions (Research desk) · How this site is sourced

Michigan's operative instrument is 'Artificial Intelligence for Attorneys: Frequently Asked Questions,' guidance the State Bar of Michigan published November 18, 2024, with fee questions added February 11, 2025. The FAQs describe themselves as neither legal advice nor an ethics opinion, and Michigan has no lawyer-facing formal opinion. Applying Michigan Rules 1.1 through 8.4, they make lawyers responsible for all AI-generated content and require citation verification, require client consent under MRPC 1.6(c)(1) before protected information is input into a tool, impose no general duty to tell clients or courts about AI use absent rule triggers, and limit billing to actual time spent, with AI subscriptions normally overhead and matter-specific per-use costs chargeable only by advance agreement.

Quick answer

  1. Instrument: 'Artificial Intelligence for Attorneys: Frequently Asked Questions' (guidance).
  2. Published by the State Bar of Michigan on November 18, 2024.
  3. Fee questions added February 11, 2025; expressly not an ethics opinion.
  4. MRPC 1.6(c)(1) consent required before inputting protected information.
  5. No general duty to disclose AI use to clients or courts absent rule triggers.
  6. Bill actual time only; subscriptions are overhead; per-use costs need advance agreement.

The official instruments

InstrumentTypeDateKey duties
Artificial Intelligence for Attorneys: Frequently Asked Questions Official guidance November 18, 2024 competence, verification, confidentiality, consent, communication, fees, supervision

Guidance that tells you exactly what it is

Michigan’s instrument opens with a self-description most trackers flatten away: the State Bar of Michigan’s “Artificial Intelligence for Attorneys: Frequently Asked Questions,” published November 18, 2024, is explicitly neither legal advice nor an ethics opinion. That candor sets the citation weight. Michigan has no lawyer-facing formal opinion on generative AI; what it has is working guidance from the state bar, organized as questions and answers and kept current, with fee questions added on February 11, 2025. The update habit is worth noticing, because a document that grows in response to member questions behaves less like a frozen pronouncement and more like a maintained manual. For the bench there is a separate, older instrument, judicial ethics opinion JI-155 from 2023 on judges’ AI competence, which says nothing about practicing lawyers but confirms Michigan has been thinking about AI on both sides of the courtroom for a while.

The FAQs’ analytical spine is broad: they run generative AI across Michigan Rules 1.1 through 8.4 rather than isolating two or three favorite duties. The premise underneath every answer is accountability. Lawyers are responsible for all AI-generated content, and citations must be verified. Nothing in the format softens that; a Q&A that ends with the lawyer owning the output is stating the same rule a formal opinion would.

The FAQs’ most operationally useful holding is the consent trigger, because it names its rule. Client consent is required under MRPC 1.6(c)(1) before protected information is input into an AI tool. Pinning the duty to a subsection converts a vague “get consent for AI” instinct into a checkable condition: is protected client information about to enter this system? If yes, consent comes first; if no, the trigger is not tripped. That is a cleaner test than most states offer, and it points intake and matter teams to the precise moment the duty attaches, the input. How that consent conversation is run is the subject of the client disclosure guide.

On disclosure more broadly, Michigan sits with the quieter states: the FAQs impose no general duty to tell clients or courts about AI use absent rule triggers. The pairing is deliberate and coherent. Michigan does not want announcements; it wants control of inputs, consent where Rule 1.6(c)(1) demands it, and verified outputs the lawyer stands behind. The comparison table shows how that posture differs from consent-first jurisdictions, which is exactly the kind of variance a multi-state policy has to absorb.

Three billing lanes, and the Michigan build

The February 11, 2025 fee additions divide AI economics into three lanes that keep getting blurred elsewhere. Lane one: only actual time spent may be billed, so hours the tool eliminated are not billable hours. Lane two: AI subscriptions are normally overhead, absorbed by the firm like any research platform license. Lane three: matter-specific, per-use AI costs can move to the client, but only by advance agreement, never retroactively. A firm that memorializes lane three in its engagement letter and keeps lanes one and two straight in its billing system has implemented the Michigan fee analysis; the billing ethics explainer shows how other instruments reach parallel results.

Assembling the whole state into a workflow: screen tools before use and identify where protected information could flow, capture MRPC 1.6(c)(1) consent before any such input, verify every citation and factual assertion in AI output before it leaves the firm, and encode all of it in a written policy with training so responsibility for AI-generated content is assigned rather than assumed; the policy template covers each piece. The courtroom adds its own layer, since our dataset records no statewide Michigan court rule on AI filings: individual judges’ standing orders can still impose certification or disclosure duties the FAQs do not, so the assigned judge’s requirements get checked before each filing, per the court orders guide. Updates to the FAQs, which have already grown once, will appear in the changelog as they are verified.

Frequently asked questions

Does Michigan have a formal AI ethics opinion for lawyers?

No lawyer-facing formal opinion exists. The State Bar of Michigan's FAQs, published November 18, 2024, state expressly that they are neither legal advice nor an ethics opinion. Michigan does have JI-155, a 2023 judicial ethics opinion on judges' AI competence, but that instrument addresses the bench, not practicing lawyers.

When do Michigan lawyers need client consent for AI use?

Before inputting protected information into an AI tool. The FAQs tie the requirement to a specific rule subsection, MRPC 1.6(c)(1), which makes the trigger unusually auditable: the consent question is answered by asking whether protected client information is about to enter the tool, not by debating AI use in the abstract.

Do Michigan lawyers have to tell clients or courts they used AI?

Not as a general matter. The FAQs recognize no general duty to inform clients or courts about AI use absent rule triggers. Lawyers remain responsible for all AI-generated content and must verify citations, so the accountability sits in the work product rather than in a disclosure statement.

How do the Michigan FAQs treat AI billing?

In three lanes, added as fee questions on February 11, 2025. Only actual time spent may be billed. AI subscriptions are normally firm overhead that cannot be passed through. Matter-specific, per-use AI costs may be charged to a client only by advance agreement.

Primary sources cited

Related guides

About the editor: MHSB Solutions, Research desk. MHSB Solutions is not a law firm. Everything on this site is legal information keyed to primary sources, not legal advice.

For licensed attorneys and firm operators. This site is legal information, not legal advice, and no attorney-client relationship is formed by using it. Rules change; verify against the primary sources linked on every page and consult a licensed attorney in your jurisdiction before acting.