EnglishEspañol
AI Meeting Recording LawsMichigan Laws flagMichigan Laws
Michigan flag

Michigan

Michigan AI Meeting Recording Laws (2026)

Independently fact-checkedEditor-reviewedLaw current as of November 1, 2026By Recording Law Editorial TeamReviewed November 2, 202613 min read

Independently fact-checked against primary sources (last audited November 1, 2026). · Reviewed by the RecordingLaw editorial team. · Law checked current as of November 1, 2026. · 3 primary sources cited on this page. How we verify our legal content

Michigan AI Meeting Recording Laws (2026)

AI meeting recorders occupy a legal gray zone in Michigan. Michigan is a one-party consent state for people who take part in a conversation: the Court of Appeals held in Sullivan v. Gray (1982) that the eavesdropping statute, on its face, excludes participant recording. Whether an AI bot qualifies as a participant or a third-party eavesdropper under the statute remains unresolved.

Michigan's eavesdropping law protects conversations from outsiders, not from the people taking part in them. The statute at Mich. Comp. Laws Section 750.539c prohibits using any device to eavesdrop on a private conversation "without the consent of all parties." But the act defines eavesdropping as capturing "the private discourse of others," and Michigan courts have held for decades that a participant in a conversation cannot "eavesdrop" on their own conversation. That makes Michigan a one-party consent state for participants, while a non-participant still needs the consent of all parties.

This participant rule matters enormously for AI meeting recording tools. When Otter.ai, Fireflies.ai, or another AI notetaker joins a video call, is the bot a "participant" in the conversation or a third-party eavesdropper? The answer determines whether the bot can legally record without everyone's consent, and no Michigan court has directly resolved this question as of November 2026.

This article covers Michigan's consent framework, the participant exception, how it applies to AI meeting tools, penalties for violations, and what employers need to know. Consult an attorney for advice specific to your situation, as this article provides general legal information and not legal advice.

The Statutory Text vs. Judicial Interpretation

Michigan's eavesdropping statute reaches, in its own words, "any person who is present or who is not present during a private conversation and who wilfully uses any device to eavesdrop upon the conversation without the consent of all parties thereto, or who knowingly aids, employs or procures another person to do the same." That consent requirement applies only to eavesdropping, so who it reaches depends on how the statute defines eavesdropping.

The participant exception does not come from that sentence. It comes from the separate definitions section. MCL 750.539a(2) defines "eavesdrop" or "eavesdropping" as "to overhear, record, amplify or transmit any part of the private discourse of others without the permission of all persons engaged in the discourse." Michigan courts have treated the words "of others" as the limiting language: someone who is part of a conversation is not capturing the private discourse of others, but taking part in their own. That reading, established in Sullivan v. Gray (Mich. Ct. App. 1982), has been the governing interpretation for over four decades.

The Sullivan court found that "the statutory language, on its face, unambiguously excludes participant recording from the definition of eavesdropping by limiting the subject conversation to 'the private discourse of others.'" A participant recording their own conversation is not intercepting the discourse of "others." Note what the exception has to work around: the phrase "who is present or who is not present" in MCL 750.539c is the strongest textual argument on the other side, and it is the phrase the exception's critics have leaned on.

Fisher v. Perron: A Federal Prediction, Not a Michigan Ruling

In March 2022, the U.S. Sixth Circuit Court of Appeals directly addressed this question in Fisher v. Perron, 30 F.4th 289 (6th Cir. 2022). The case involved siblings in a dispute over their late mother's estate. Michelle Perron recorded phone conversations with her brother Frank Fisher and other siblings without their knowledge, then attempted to use those recordings in estate proceedings.

The Sixth Circuit rejected Fisher's claim, agreeing with the district court that "under the current law in Michigan, a participant does not violate Michigan's eavesdropping statute by recording a conversation without the consent of the other participants."

The court was explicit about the limits of what it was doing. It stated that "the Michigan Supreme Court has yet to determine whether this statutory language allows for secret participant recordings or requires the consent of all parties to record a call," and that the panel therefore "must predict how the Michigan Supreme Court would rule based on 'all the available data.'" That is an Erie prediction of state law by a federal court. It resolved the case in front of it, but it does not bind any Michigan court and it did not settle the question as a matter of Michigan law. For an article about a question a Michigan court has not yet reached, that distinction matters.

The Michigan Supreme Court Has Not Answered the Question

The participant exception has been challenged, and the challenge was serious. Sullivan v. Gray itself drew a dissent. In 2019, a federal district court in AFT Michigan v. Project Veritas, 397 F. Supp. 3d 981 (E.D. Mich. 2019), worked through the same "available data" and reached the opposite conclusion, holding that the Michigan Supreme Court would read the statute to require the consent of all parties. That court reasoned that Sullivan's construction "contravenes the Legislature's intent made clear" by the statutory phrase covering "any person who is present." It then certified the question to the Michigan Supreme Court.

The Michigan Supreme Court declined in May 2021 to answer it. In re Certified Question, 959 N.W.2d 172 (Mich. 2021). A declination is not an endorsement: the court simply did not take the question. In November 2021 the same district court granted reconsideration and reversed itself, holding that "the statute is not violated when a conversation is recorded by one of its participants," which is why there is no live contrary holding today.

So the accurate statement as of November 2026 is narrower than "settled." Every currently operative decision points the same way, and the Legislature has not amended the statute in six decades of judicial interpretation. But the state's highest court has never ruled, one federal court reached the opposite result before withdrawing it, and the textual argument against the exception has never been answered by a Michigan court of last resort. Anyone planning around the exception should treat it as the prevailing interpretation, not as a closed question.

A Sullivan Aside: Third-Party Agents

Sullivan v. Gray also said, in passing, that "while a participant may record a conversation, he apparently may not employ third parties to do so for him," because the other speakers cannot gauge what they say around a listener they do not know about. That remark was not the question the court decided, but it is directly relevant to AI meeting tools. If an AI bot is classified as a third-party agent rather than a participant, it could fall outside the participant exception and require all-party consent.

Empty corporate boardroom with leather chairs around a long table, a dark blank wall screen at the end

How Michigan's Participant Rule Applies to AI Meeting Bots

The Central Question: Participant or Eavesdropper?

Michigan's participant exception creates a binary legal test for AI meeting recording tools:

Michigan law has not decided whether an AI meeting bot is a participant or a third-party eavesdropper, so getting every participant’s consent is the safest approach.
Until a Michigan court rules on AI note-takers, announce the bot and get everyone’s agreement, especially if the vendor uses recordings itself.

If the AI bot is a "participant": It can record the conversation without the consent of other parties, just like any human participant. Under this classification, tools like Otter.ai and Fireflies.ai would be legal in Michigan as long as the account holder (who is a participant) authorized the recording.

If the AI bot is a "third-party eavesdropper": It requires consent from all parties to the conversation. Under this classification, an AI bot that joins a meeting and records without everyone's knowledge would violate MCL 750.539c, and the person who deployed it could also be liable for procuring an illegal eavesdrop.

No Michigan court has directly ruled on this question. The answer likely depends on several factors.

Factor 1: How the Bot Joins the Meeting

An AI bot that joins a meeting as a visible participant with its own name displayed (such as "Otter Notetaker") has a stronger argument for participant status than a bot that operates invisibly in the background. However, "joining" a meeting may not be enough. Sullivan v. Gray said a potential eavesdropper must be "a third party not otherwise involved in the conversation being eavesdropped on," and noted that one may be "present" during a conversation without being a party to it. The opinion does not further define who counts as a participant.

Factor 2: Whether the Bot Acts Independently

Sullivan's remark about third parties is significant here. If an AI bot merely records on behalf of the meeting host (acting as a tool or extension of the host), it might be treated as the host's recording device rather than a separate entity. But if the bot processes recordings independently, sends them to the vendor's servers for AI training, or generates outputs that the host did not direct, courts may view it as an independent third party.

Factor 3: The Ambriz v. Google "Capability Test"

In Ambriz v. Google (N.D. Cal., Feb. 10, 2025), a federal judge refused to dismiss California wiretapping claims over Google's contact-center AI. On whether a software vendor is a third-party eavesdropper under California Penal Code 631, the court noted that trial courts are split: some require that the vendor actually use the recordings for its own purposes, while others ask only whether it is capable of doing so. The court applied the broader capability approach, first used in Javier v. Assurance IQ (N.D. Cal. 2023). It was a pleading-stage ruling that bound no other court, and in October 2026 the same judge dismissed the remaining plaintiffs' claims on consent and intent grounds (with leave to amend) and sent others to arbitration.

No Michigan court has considered whether that reasoning carries over to MCL 750.539c, which asks whether the recorder is "a third party not otherwise involved in the conversation" (Sullivan v. Gray). A Michigan court could look to it, but none has, so treat third-party classification of AI bots as an open risk rather than a prediction.

Factor 4: The Otter.ai Litigation

In re Otter.AI Privacy Litigation (N.D. Cal., No. 5:25-cv-06911, formerly Brewer v. Otter.ai) consolidates four class actions filed against Otter.ai in 2025. The plaintiffs allege that Otter.ai's Notetaker joins meetings without obtaining consent from all participants, records conversations of non-account holders, and uses the data to train machine learning models.

In August 2026 the court let the core federal Wiretap Act, California privacy and Illinois biometric claims proceed past a motion to dismiss. The federal claim survived on the theory that using meeting conversations without consent to train Otter's own AI models could be a tortious purpose that defeats the federal one-party consent exception, even if Otter counted as a party. The case is now in discovery, and no court has decided whether Otter actually violated the law. Only the federal claim would reach a Michigan meeting directly; the California and Illinois claims rest on those states' own statutes, and nothing in the case construes Michigan's eavesdropping act.

ToolDefault BehaviorMichigan Risk Assessment
Otter.aiBot joins automatically via calendar integration; host controls consent settingsUnresolved: no court has classified it under Michigan law
Fireflies.aiBot joins as named participant; Zoom may show recording pop-upUnresolved: no court has classified it under Michigan law
Zoom AI CompanionBuilt-in feature; shows recording indicatorUnresolved: no court has decided whether built-in features are treated differently
Microsoft CopilotIntegrated into Teams; transcription notification displayedUnresolved: no court has decided whether built-in features are treated differently
Google Gemini in MeetBuilt-in summarization; recording banner shownUnresolved: no court has decided whether built-in features are treated differently
Read.aiBot joins meetings; sends consent requestsUnresolved: consent from every participant avoids the classification question

No court has decided whether a built-in feature such as Zoom AI Companion, Microsoft Copilot or Google Gemini is treated differently from a separate bot under Michigan law. Ambriz v. Google involved a vendor's own AI built into businesses' call systems, and that court still found a plausible claim that the vendor was a third party. What matters is whether the vendor keeps or uses the recording for its own purposes, so check each vendor's data terms and announce the recording either way.

Two monitors on a desk arm showing soft abstract glows with a keyboard and mouse in front

Penalties for Violating Michigan Eavesdropping Laws

Criminal Penalties

OffenseClassificationMaximum PrisonMaximum Fine
Eavesdropping on a private conversation, MCL 750.539cFelony2 years$2,000
Installing or using a device in a private place, first offense, MCL 750.539d(3)(a)(i)Felony2 years$2,000
Installing or using a device in a private place, after a prior conviction, MCL 750.539d(3)(a)(ii)Felony5 years$5,000
Distributing or transmitting a recording obtained in violation of MCL 750.539d, under MCL 750.539d(3)(b)Felony5 years$5,000
Using or divulging information known to have been obtained in violation of MCL 750.539b, 539c, or 539d, under MCL 750.539eFelony2 years$2,000

Under MCL 750.539c, violations are classified as felonies. The device offenses in MCL 750.539d escalate: the 2-year exposure is the first-offense tier, and both a repeat violation and distributing the resulting recording carry up to 5 years and $5,000. Both the person who deploys the AI recording tool and the vendor whose tool performs the recording could potentially face liability.

Civil Remedies

MCL 750.539h provides a private right of action for victims of illegal eavesdropping. A person whose private conversation is illegally recorded may sue for an injunction against further eavesdropping, actual damages, and punitive damages as determined by the court or jury. In the context of AI meeting tools, each recorded participant who did not consent could potentially bring a separate civil claim.

Evidentiary Impact

Michigan's eavesdropping act contains no suppression rule. Sections 750.539a through 750.539i create criminal penalties and the civil remedies in MCL 750.539h, but none of them directs a court to exclude a recording made in violation of the act. MCL 750.539i points the other way entirely: it provides that proof of the installation of such a device in a private place is prima facie evidence of a violation of MCL 750.539d, which is a presumption against the person who installed it rather than a bar on the recording.

Whether a particular recording comes into evidence therefore turns on the ordinary rules of evidence and on the judge hearing the matter, not on an automatic statutory exclusion. Do not assume an AI meeting transcript will be kept out of a proceeding because it was captured without consent, and do not assume it will be let in.

The Bullard-Plawecki Act

Michigan's Bullard-Plawecki Employee Right to Know Act adds an additional layer of protection in the employment context. The Act prohibits employers from gathering or maintaining records of employees' non-employment activities, associations, or political activities without consent. AI meeting tools that capture conversations beyond their intended business scope could implicate this statute.

Employer and Workplace Recording Rules

The Workplace Recording Dilemma

Michigan employers face a particular challenge with AI meeting recording. While the participant exception allows an employee who is part of a conversation to record it, deploying an AI bot to record on the employer's behalf raises Sullivan's concern about third parties recording for a participant.

Employers should not assume that because Michigan is effectively a one-party consent state for participants, they can freely deploy AI recording tools without disclosure. The prudent approach is to treat AI bots as potential third-party eavesdroppers and obtain consent from all meeting participants.

A black round conference speakerphone on a wooden meeting table with colleagues seated around it

Best Practices for Michigan Employers

Employers using AI meeting tools in Michigan should consider several compliance measures. Written policies should clearly state that AI recording tools may be used in workplace meetings. Verbal announcements at the start of each recorded meeting provide real-time notice. Meeting invitations should note that AI transcription will be active. Opt-out mechanisms should be available for participants who do not wish to be recorded.

Video Surveillance Considerations

Michigan law permits employers to use video surveillance in the workplace for legitimate business purposes, but cameras may not be installed in areas where employees have a reasonable expectation of privacy (such as restrooms, locker rooms, and changing areas). Audio recording through workplace surveillance systems is subject to the eavesdropping statute and its participant exception.

Remote and Multi-State Meetings

When Michigan employees participate in meetings with colleagues in other states, courts have not settled which state's law governs. Some courts have applied the law of the place where the recording was made (Commonwealth v. Larrison in Pennsylvania; Ditech Financial v. Buckles in Nevada, 2017), while the California Supreme Court in Kearney v. Salomon Smith Barney (2006) applied California law only prospectively to calls recorded out of state with California residents, and applied Georgia law to damages for the past recordings. If any participant is in an all-party consent state like Massachusetts, California, or Illinois, notifying everyone is the safe practice. Michigan employers with distributed workforces should implement a universal disclosure policy requiring announcement of AI recording at the start of every meeting.

Federal Law and Michigan's Framework

18 U.S.C. Section 2511

The federal Wiretap Act follows a one-party consent standard. A person who is a party to a communication, or who has one party's consent, may record it under federal law unless the recording is made for the purpose of committing a criminal or tortious act (18 U.S.C. 2511(2)(d)). That limit is why the federal claim against Otter.ai survived dismissal in August 2026. Federal penalties for violations include up to 5 years in prison and fines up to $250,000.

Michigan's eavesdropping statute and the federal Wiretap Act can apply simultaneously to the same recording. Each applies on its own terms, so a recording has to satisfy both. For AI meeting recording in Michigan, the participant exception creates functional one-party consent for human participants, but the unresolved question of AI bot classification means the state standard could be stricter than federal law for AI tools.

Cross-Border Implications

AI meeting recordings frequently involve participants in multiple states. A Michigan-based company recording a meeting with participants in California (all-party consent) or Illinois (which has its own Biometric Information Privacy Act) cannot be sure which state's law a court would apply, because courts have not settled that question. The safest approach is to adopt all-party consent practices regardless of Michigan's participant exception.

Looking Ahead: Michigan's AI Recording Landscape

Michigan's eavesdropping statute was written long before AI meeting assistants existed. The participant exception, crafted in 1982 for human conversations, does not map cleanly onto AI bots that join meetings, process audio through machine learning models, and store data on vendor servers.

Several developments could resolve the open question about AI bots. The Michigan Legislature could amend MCL 750.539c to explicitly address AI recording tools and clarify whether they fall within or outside the participant exception. Michigan courts could rule directly on whether an AI bot qualifies as a participant or a third-party eavesdropper. Federal decisions such as In re Otter.AI Privacy Litigation and Ambriz v. Google do not bind Michigan courts, but a Michigan court could find their reasoning persuasive.

An open metal filing cabinet drawer full of green hanging folders with blank tabs, window light on the left

Until one of these developments occurs, businesses and individuals using AI meeting recording tools in Michigan should err on the side of disclosure and consent. The open question itself is a risk factor, as it means any recording made without all-party consent could be challenged if a court later determines that AI bots fall outside the participant exception.

More Michigan Laws

Explore other Michigan law topics on Recording Law:

More Michigan Laws

Frequently Asked Questions

Is it legal to use an AI meeting recorder like Otter.ai in Michigan?

For the AI bot, the answer is unsettled. Michigan is a one-party consent state for participants: MCL 750.539c bars eavesdropping without the consent of all parties, but Sullivan v. Gray (1982) held that the definition of eavesdropping in MCL 750.539a(2), which covers only 'the private discourse of others,' on its face excludes participant recording. The Sixth Circuit predicted the same result in Fisher v. Perron (2022) while noting that the Michigan Supreme Court has never decided the question. Whether an AI bot counts as a participant or a third-party eavesdropper has not been decided by any Michigan court as of November 2026.

Do I need to disclose that an AI notetaker is recording a meeting in Michigan?

It is not strictly required for a human participant recording their own conversation, under the participant exception established in Sullivan v. Gray. However, because it is unresolved whether the AI bot itself counts as a participant or as a third party the host directed to record, disclosure is generally the safer practice. Sullivan also said, in passing, that a participant 'apparently may not employ third parties' to record for them, which is why a bot acting as a separate third party could fall outside the exception.

What is the difference between an AI bot being a participant versus a third-party eavesdropper under Michigan law?

If a court treats the AI bot as a participant, it may record without the other parties' consent, similar to a human participant relying on the Sullivan exception. If a court instead treats the bot as a third-party eavesdropper, all-party consent would be required. A California federal court in Ambriz v. Google applied a 'capability' test to treat an AI vendor as a possible third party under California law, but no Michigan court has adopted that reasoning. Relevant factors discussed include how visibly the bot joins the meeting and whether it processes or stores data independently of the person who activated it.

What are the penalties for illegal eavesdropping in Michigan?

Violations of MCL 750.539c are felonies, punishable by up to two years in prison and a fine of up to $2,000. The related device offenses in MCL 750.539d run higher: a repeat violation, or distributing a recording obtained through one, carries up to five years and a $5,000 fine. MCL 750.539h also provides a private right of action, letting a person whose private conversation was illegally recorded sue for an injunction, actual damages, and punitive damages. The act contains no suppression rule, so whether an unlawfully made recording is admissible is decided under the ordinary rules of evidence rather than by an automatic statutory bar.

Has a Michigan court ruled on whether AI meeting bots are participants or eavesdroppers?

No. As of November 2026, no Michigan court has directly addressed whether an AI meeting bot qualifies as a participant or a third-party eavesdropper under MCL 750.539c. Until that question is resolved, businesses and individuals using AI meeting tools in Michigan are generally advised to err on the side of disclosure and all-party consent.

Updates

We corrected the description of Kearney v. Salomon Smith Barney (California law applied only going forward; Georgia law governed damages for past calls), relabeled Sullivan v. Gray's remark about third-party agents as an aside, and reworded how the page describes the eavesdropping statute's consent language.

We corrected this page to state plainly that Michigan is a one-party consent state for participants under Sullivan v. Gray, removed an unsupported prediction and vendor risk grades based on a California ruling, replaced a 'strictest law governs' rule with the unsettled conflict-of-laws picture, added the federal criminal-or-tortious-purpose limit, and updated the Otter.ai and Google cases to their 2026 rulings.

Independently fact-checked against the cited primary sources; governing law re-checked for recent changes

Corrected the description of Sullivan v. Gray, which did not define a participant by active communication, to what the Michigan Court of Appeals actually said about third-party eavesdroppers.

Corrected the quotation of Michigan’s eavesdropping statute and its penalty ranges, clarified that Fisher v. Perron was a federal prediction rather than a binding Michigan ruling, added the AFT Michigan challenge and the Michigan Supreme Court’s declination to answer it, and removed an unsupported claim that illegally obtained recordings are generally inadmissible.

Independently fact-checked against the cited primary sources; governing law re-checked for recent changes

Corrected the civil remedies listed under MCL 750.539h: the statute authorizes an injunction, actual damages, and punitive damages, not attorney's fees, which the article had incorrectly listed while omitting the actual punitive-damages remedy.

Governing law re-checked for recent changes

Reviewed and approved by an editor

Sources and References

  1. Mich. Comp. Laws Section 750.539c - Eavesdropping(legislature.mi.gov).gov
  2. Sullivan v. Gray, 117 Mich. App. 476 (1982) - Participant Exception(law.justia.com)
  3. Fisher v. Perron, 30 F.4th 289 (6th Cir. 2022) - One-Party Consent Affirmed(butzel.com)
  4. In re Otter.AI Privacy Litigation, N.D. Cal., No. 5:25-cv-06911 (docket; formerly Brewer v. Otter.ai)(courtlistener.com)
  5. Ambriz v. Google - AI Wiretapping Ruling (Courthouse News, 2025)(courthousenews.com)
  6. Michigan Bullard-Plawecki Employee Right to Know Act(legislature.mi.gov).gov
  7. 18 U.S.C. Section 2511 - Federal Wiretap Act(law.cornell.edu)
  8. MCL 750.539h - Civil Remedies for Eavesdropping(legislature.mi.gov).gov
  9. MCL 750.539a - Definitions, including eavesdrop as the private discourse of others(legislature.mi.gov)
  10. MCL 750.539d - Devices in a private place, penalty tiers up to 5 years and $5,000(legislature.mi.gov)
  11. MCL 750.539e - Using or divulging information obtained in violation of the act(legislature.mi.gov)
  12. MCL 750.539i - Prima facie evidence provision; the act contains no suppression rule(legislature.mi.gov)
  13. Fisher v. Perron, 30 F.4th 289 (6th Cir. 2022) - published opinion(opn.ca6.uscourts.gov)
Share: