As Michigan embarks on a new election year, the state’s lawmakers concluded their final session before adjourning for the year. The legislative agenda included bills aimed at safeguarding poll workers and regulating the use of artificial intelligence in social advertising.
House Bills 5141, 5143, and 5144, designed to mandate disclaimers in social ads created using artificial intelligence and impose penalties for attempts to deceive voters through such ads, are now awaiting the governor’s approval. These new election law violations encompass actions like intimidating or obstructing election officials from fulfilling their duties.
Following approval in the House last year, House Bill 4129 successfully passed through the Senate without amendments. However, before reaching the House floor for a vote, Senate Bills 5141, 5143, 554, and 5145 underwent further revisions.
A U.S. Senate panel was briefed by authorities on the exodus of state and local election personnel due to harassment.
Under House Bill 4129, introduced by State Representative Kara Hope (D-Holland), it would be considered a misdemeanor to intimidate or hinder election officials in the performance of their responsibilities, punishable by up to 93 days in jail and a $500 fine.
A repeat offense could lead to a $1,000 fine and a potential one-year imprisonment. Subsequent violations, including the second offense, would escalate to criminal charges.
House Bill 5141 mandates that any individual, group, or entity disseminating qualifying social ads must include a disclaimer indicating the use of artificial intelligence in generating the ad, either wholly or partially.
Failure to comply with the disclaimer requirements, as outlined in the House-approved version, could result in a misdemeanor punishable by up to 93 days in jail, a fine of up to \(1,000, or both. For a second offense, fines could increase to \)1,500. Subsequent violations could incur penalties of up to $2,000, two years of imprisonment, or both.
However, the Senate’s final version, endorsed by the House, significantly reduced the penalties, categorizing the offenses as civil infractions. Initial violations carry a maximum fine of \(250, with each subsequent offense incurring an additional \)1,000 fine. Exceptions are made for movie or comedy ads and media content, where each ad broadcast is treated as a separate violation.
The final iteration of the bill even removed provisions penalizing individuals circulating AI-generated political content without disclaimers outside official campaigns.
While the legislators have eased penalties for disseminating highly deceptive media, commonly known as deepfakes, in an attempt to mislead voters, the core principles remain unchanged.
Individuals failing to include disclaimers on potentially misleading ads risk misdemeanor charges, fines of up to \(500, and a maximum of 90 days in prison for the first offense. Subsequent infractions within five years could result in misdemeanor charges, up to 5 years of imprisonment, and an additional \)1,000 fine.
Furthermore, HB 5143’s definition of AI was revised by the government to describe it as “a machine-based system that generates predictions, recommendations, or decisions affecting real or virtual environments based on human-defined objectives, utilizing machine and human inputs to perceive real-world and digital environments, automatically abstract perceptions into models, and infer knowledge or action options.”
State Representative Matthew Bierlein (R-Vassar), a sponsor of the AI legislation, emphasized the rapid evolution of AI technology and its significant impact on elections and political information consumption.
Bierlein stressed the importance of establishing a legal framework to address AI’s effects and risks promptly, underscoring the bills’ role in safeguarding election integrity and citizens’ interests. Failure to establish clear guidelines now could lead to challenges in the future.
Michigan Secretary of State Benson highlighted the importance of enhancing fines for individuals threatening or harassing poll workers to ensure the safety of election officials statewide.
Benson also expressed satisfaction with the legislation addressing the influence of artificial intelligence on public discourse and election security, emphasizing the need for collaborative efforts at state and federal levels to protect citizens from potential interference by foreign or domestic entities using AI to manipulate election processes.
