Michigan AI Legislation in 2026

The Michigan Picture

Michigan has chosen a targeted, harm-specific path on artificial intelligence rather than a single comprehensive statute, and the clearest expression of that choice is its election law. In late 2023 the Legislature enacted a package, HB 5141, HB 5143, and HB 5144, effective February 13, 2024, that made Michigan one of the first handful of states to regulate AI in political advertising. The framework does two things at once. It requires a clear and conspicuous disclosure whenever a political advertisement is generated in whole or substantially by AI, and it criminalizes the distribution of materially deceptive AI-manipulated media of a candidate within 90 days of an election. This pairing of a civil disclosure mandate with a criminal deception offense is the distinctive Michigan model.

The second pillar arrived in 2025 with the intimate deepfake package, HB 4047 (Public Act 11 of 2025) and its criminal companion HB 4048, signed by Governor Whitmer on August 26, 2025. Here Michigan created a civil cause of action and parallel criminal penalties for nonconsensual sexually explicit synthetic media depicting an identifiable person, with felony exposure for aggravating conduct and protections allowing victims to litigate under a pseudonym. Taken together, the election and deepfake measures show a state that legislates against concrete, demonstrable harms, deceptive campaigns and synthetic sexual abuse, rather than attempting to regulate AI systems or developers in the abstract.

What this means in practice is that I would not advise builders to expect a Colorado-style or EU-style risk classification regime in Michigan today. There is no omnibus AI law, no algorithmic discrimination statute with enacted teeth, and no general developer duty of care on the books as of June 2026. Healthcare AI is being shaped administratively through the Health Information Technology Commission and its AI Task Force inside the Department of Health and Human Services, which signals where future obligations may originate but does not yet impose statutory duties. Some 2025-2026 session proposals on government AI use and AI safety remain pending, not enacted, and should be tracked rather than treated as binding.

For anyone deploying AI in Michigan, the compliance priorities are concrete. If you touch political or electoral content, you must implement the AI disclosure language and timing rules and avoid any deceptive candidate media inside the 90-day pre-election window. If your product can generate or host imagery of real people, you must treat nonconsensual sexual synthetic media as both a civil and criminal liability surface and build provenance, consent, and takedown mechanisms accordingly. Everything else in Michigan remains governed by general law, so the smart move is to align now with the strictest applicable comparable state regime and monitor the session and the health task force for the next wave.