Michigan: AI Ethics Guidance for Law Firms

Bottom line for a 5-50 attorney Michigan firm: The State Bar’s AI FAQs are the main practice guidance and are expressly not an ethics opinion; State Bar ethics opinions such as RI-381 (2020), which the FAQs cite on vendor security, are “advisory and non-binding in nature.” The FAQs say a lawyer “must review” AI-generated information “to confirm that all information and cites are correct,” and must obtain client consent under MRPC 1.6(c)(1) before entering protected information into an AI program. Federal judges in the Eastern District have sanctioned AI-fabricated citations in Seither & Cherry (2025) and Fay v. Warren Hospitality Suites (2026).

Start with: a written AI use policy, a citation verification checklist for every filing, and vendor due diligence records that follow RI-381.

State Bar of Michigan AI FAQs for Attorneys

Citation: State Bar of Michigan, Artificial Intelligence for Attorneys: Frequently Asked Questions (published 2024-11-18; a question on charging clients for developing AI tools added 2025-02-11). Primary source.

Status: Non-binding practice guidance, not a numbered ethics opinion. The FAQs state that they “are neither legal advice nor an ethics opinion, and are not a substitute for your obligation to adhere to the requirements of the Michigan Rules of Professional Conduct,” and that they do not reflect “the ethical implications of any updates, modifications, or added features.”

What it says

• Competence and verification (MRPC 1.1). Lawyers must understand AI’s capabilities and limits, including hallucination risk, and “must review all information generated to confirm that all information and cites are correct.” The FAQs call this “especially important when any document is filed with a court or tribunal to ensure a lawyer does not violate MRPC 3.3.”
• Responsibility for output. “Lawyers are responsible for all legal documentation whether drafted by their own hand or through a technological program,” as with documentation drafted by staff (citing MRPC 5.1 and 5.3).
• Confidentiality (MRPC 1.6). “If it is necessary to input any protected information, the lawyer must first receive the client’s consent under MRPC 1.6(c)(1).” The lawyer must also “take reasonable steps to determine whether the AI provider has adequate security measures,” citing RI-381.
• Billing time (MRPC 1.5). Where AI shortens a task, the lawyer may charge “for the actual time to review the motion, validate citations, and otherwise exercise independent professional judgment.”
• AI costs. “In the absence of disclosure to a client in advance of the engagement, the overhead expense of using the AI tool should be absorbed by the lawyer or law firm.” A per-use expense may be billed “provided that expense was agreed to at the time of engagement.” Lawyers “should not charge a client for expenses related to learning how to use an AI tool or service.”
• Telling clients and courts. The Rules “do not currently create a duty to inform the client that a lawyer is using an artificial intelligence program, except when use of the program implicates the rules of professional conduct”; for example, client consent is required if use may expose confidential information. No duty to disclose AI use to a court arises unless the court requests it, including under local rules or a case management order.

Notable gaps

The FAQs do not address AI labeling in filings beyond pointing to court-specific requirements, and they do not address multi-jurisdictional practice where another state’s rules are stricter.

Ethics Opinion RI-381 (cybersecurity foundation for AI vendor diligence)

Citation: State Bar of Michigan, Ethics Opinion RI-381 (February 21, 2020).

Status: Advisory ethics opinion. The State Bar’s opinion index states that its “ethics opinions are advisory and non-binding in nature.” RI-381 predates generative AI and does not mention it; the AI FAQs cite it in their confidentiality answer. It concludes that “a lawyer must conduct appropriate due diligence to ensure the vendor’s data security controls will reasonably protect against cyber risks,” and addresses cybersecurity policies, periodic reassessment, and notice to clients of material breaches.

Ethics Opinion JI-155 (judicial AI competence)

Citation: State Bar of Michigan, Ethics Opinion JI-155 (October 27, 2023).

Status: Advisory opinion addressed to judicial officers. It states: “Judicial officers, like lawyers, have an ethical obligation to maintain competence with and further educate themselves on advancing technology, including but not limited to artificial intelligence (AI).” It notes that the Code of Judicial Conduct could be implicated “if the tool’s algorithm or training data creates bias,” and that “Judicial use of AI must distinguish between using an AI application to decide and using AI to inform a decision.”

State Bar AI Workgroup Report (June 2025)

On June 11, 2025, the State Bar announced Transforming the Legal Profession in the Age of AI, authored by the Board of Commissioners’ AI Workgroup and “intended to be a living document.” Report PDF. It is not an ethics opinion. It discusses MRPC 1.1 and AI’s implications for unauthorized practice under MCL 600.916, and makes five recommendations: keeping member information on AI current, keeping State Bar staff informed, collaborating with law schools and other bar organizations, providing reliable public information on AI used in place of representation, and having State Bar committees monitor AI and advise the Board.

What federal courts in Michigan require for AI use in filings

No district-wide AI standing order or local rule is in force in either Michigan federal district as of 2026-09-30. The Eastern District’s judges approved proposed Local Rule 5.1(a)(4) for publication on December 4, 2023 (notice, comments due January 19, 2024). It would have required disclosure of generative AI use and human verification of cited authority. It was not adopted: the current E.D. Mich. LR 5.1 has no AI provision. The Western District’s revised local rules, adopted by Administrative Order 26-RL-067 (August 13, 2026), also contain none. Fed. R. Civ. P. 11 applies regardless, and individual judges’ orders are worth checking at the start of each matter.

In bankruptcy practice, the U.S. Bankruptcy Court for the Eastern District of Michigan posted an advisory notice regarding pleadings using generative AI on May 26, 2026. It reminds counsel and parties that they “have an independent duty and obligation to verify the accuracy of pleadings and documents prior to the filing with this Court,” treats every filing as a certification under Bankruptcy Rule 9011(b), and warns that a violation could bring sanctions. It imposes no disclosure or AI-specific certification requirement.

What Michigan state courts require for AI use in filings

Neither the Michigan Supreme Court nor any state trial court has adopted a rule requiring AI disclosure in filings; the Michigan Court Rules, as updated September 2, 2026, do not mention artificial intelligence. AI-related conduct is measured against the existing signature certification in MCR 1.109(E)(5), with sanctions under MCR 1.109(E)(6), and, for discovery papers, MCR 2.302(G).

The Michigan Judicial Council’s Generative AI and the Courts Workgroup Report (October 2024) recommended training, pilot programs, and feedback processes for GenAI tools in the courts. It states that its recommendations “do not represent the official position or policies of the Michigan Supreme Court or State Court Administrative Office.”

The Michigan Court of Appeals has addressed AI-related citation problems in individual appeals, including Groulx v. CSL (December 19, 2025) and Barber v. Morawa (June 17, 2026).

Has anyone been sanctioned for AI use in Michigan federal court?

Seither & Cherry Quad Cities v. Oakland Automation (E.D. Mich. 2025)

Citation: Seither & Cherry Quad Cities, Inc. v. Oakland Automation, LLC, No. 23-11310, consolidated with AP Electric, Inc. v. Oakland Automation, LLC, No. 4:23-cv-11342-FKB-EAS, ECF No. 92 (E.D. Mich. July 28, 2025) (Behm, J.) (consolidated opinion and order). Order. Tracker entry.

Plaintiffs’ counsel submitted briefs containing AI-generated citations with real case names but fabricated quotations and parentheticals. Judge F. Kay Behm imposed Fed. R. Civ. P. 11 sanctions of $1,485.00, which “Plaintiffs’ counsel shall pay … directly to defense counsel,” and held that “Plaintiffs’ counsel – and not Plaintiffs – are responsible for paying this penalty.” Quoting Sanders v. United States, 176 Fed. Cl. 163, 169 (2025), the order noted that generative AI programs “are known to ‘hallucinate’ nonexistent cases, and with the advent of AI, courts have seen a rash of cases in which both counsel and pro se litigants have cited such fake, hallucinated cases in their briefs.”

Other Eastern District orders

Earlier, in Loyer v. Wayne County (March 21, 2025), the court addressed AI-generated citations in a filing. More recently, in Fay v. Warren Hospitality Suites (September 9, 2026), the court imposed a $10,000 fine jointly on plaintiff’s attorney and the attorney’s firm.

What AI-related rules are pending in Michigan?

No numbered State Bar ethics opinion on attorney AI use has issued; as of 2026-09-30 the most recent opinions on the opinion index are RI-396 and JI-160 (September 2026), neither of which concerns AI. No court rule proposal on AI has been published. Two 2025 bills touch AI but not attorney practice: HB 4667 would criminalize use of AI to commit certain crimes (referred to the House Judiciary Committee June 24, 2025), and HB 4668 would set safety standards for developers of certain AI models (re-referred to the Committee on Communications and Technology March 19, 2026).

How do malpractice carriers in Michigan treat AI use?

No Michigan malpractice carrier has published AI-specific underwriting guidance in the materials reviewed for this entry, and no filed AI exclusion on an admitted-market LPL form has been located. Documented evaluation of an AI tool’s data practices, following RI-381, is the kind of record a firm would want if an AI vendor’s security failure led to a claim. Whether a given policy responds to AI-related sanctions or errors depends on its terms, which is a question for the firm’s broker.

What does my Michigan malpractice carrier ask about AI at renewal?

No Michigan carrier has published AI-specific application items in the materials reviewed for this entry. Firms that want to be ready for questions may wish to have on hand a written AI use policy, vendor due diligence records following RI-381, citation verification logs for filings, and a billing practice consistent with the FAQs’ fee guidance. Confirm application items directly with the broker.

What documentation should a Michigan firm keep on file?

Month one (foundational)

1. (Owner: managing partner + firm administrator) Written AI use policy. Cover approved tools, what categories of client data may or may not be input, the verification workflow before any AI output is filed or sent to a client, and supervision under MRPC 5.1 and 5.3.
2. (Owner: litigation lead) Citation verification checklist requiring human verification of every citation in any filing, with the verifying attorney’s initials before submission. This tracks the FAQs’ review guidance and the Rule 11 orders above.
3. (Owner: matter lead attorney) Judge-specific AI order check at the opening of each federal matter, since no district-wide rule is in force.

Months two and three (operational documentation)

1. (Owner: firm administrator + outside IT) Vendor due diligence records for each AI tool: data handling, privacy policy, terms of service, and security adequacy, following RI-381. These document the diligence RI-381 describes.
2. (Owner: managing partner + billing partner) Engagement letter AI clause. Address whether the firm uses AI tools in the representation, how AI costs are handled (the FAQs tie overhead to advance disclosure and per-use charges to agreement at engagement), and client consent before protected information is input under MRPC 1.6(c)(1).
3. (Owner: firm administrator) Training records showing attorneys and staff have current AI competence training under MRPC 1.1.

Months four to six (per-matter discipline)

These are recurring practices, not one-time projects: each new matter exercises them.

1. (Owner: billing partner) Fee billing protocol documentation. Bill the actual time spent when AI tools are used, treat AI tool costs consistently with what was disclosed or agreed at engagement, and do not charge clients for learning to use AI tools. This follows the FAQs’ fee guidance (November 2024, with a February 2025 addition).
2. (Owner: managing partner) Annual professional liability policy review documenting review of the firm’s malpractice policy for AI-related exclusions, sublimits on AI-related claims, and whether “professional services” is defined to cover AI-assisted work.

We are building practitioner resources for firms working through this list. The newsletter covers new resources as they ship.