United States - Michigan - AI in Political Advertising (PA 263-266)

Overview

The Michigan legislative package, initially introduced as House Bills 5141 through 5145 in 2023, represents a significant legislative effort to address the burgeoning challenges posed by artificial intelligence (AI) and manipulated media in the realm of political advertising. These bills have since been enacted into law as Public Acts 263-266 of 2023, marking Michigan as one of the pioneering states to establish comprehensive regulations in this critical area. The core motivation behind this legislative suite stems from growing concerns that AI-generated content, particularly 'deepfakes,' could be used to mislead voters and undermine the integrity of elections, given the increasing sophistication and indistinguishability of such media from authentic content.

The collective aim of these Public Acts is multifaceted: to mandate transparency regarding the use of AI in political advertisements, to prohibit the dissemination of materially deceptive media with electoral intent, and to establish clear penalties for violations. By amending existing statutes such as the Michigan Campaign Finance Act and the Michigan Election Law, alongside the Code of Criminal Procedure, the state has created a robust legal framework designed to safeguard democratic processes against technological manipulation. This proactive approach reflects a recognition of the rapid advancements in AI and the potential for its misuse, seeking to ensure that voters receive accurate information and are not unduly influenced by deceptive digital content during critical election periods.

Definitions

Central to the understanding and application of these Michigan Public Acts are several key definitions, meticulously crafted to delineate the scope of the regulations. 'Artificial intelligence' is defined as a machine-based system capable of making predictions, recommendations, or decisions that influence real or virtual environments, based on human-defined objectives. This definition further specifies that such systems utilize machine and human-based inputs to perceive environments, abstract these perceptions into models through automated analysis, and subsequently use model inference to formulate options for information or action. This comprehensive definition, primarily established by House Bill 5143 (now Public Act 264 of 2023), ensures that a broad range of AI applications in political messaging falls under the purview of the new regulations.

Another crucial term is 'materially deceptive media,' often colloquially referred to as 'deepfakes.' House Bill 5144 (now Public Act 265 of 2023) defines this as any image, audio, or video that falsely depicts an individual engaging in speech, conduct, or appearance in which they did not actually participate. A critical component of this definition is the requirement that a 'reasonable viewer or listener' would incorrectly believe the depicted individual genuinely engaged in the misrepresented speech or conduct. This standard aims to distinguish between satirical or clearly fictional content and media genuinely intended to deceive. The legislation also introduces the concept of a 'qualified political advertisement,' which encompasses any advertisement or sponsored content related to federal, state, or local elections or ballot questions in Michigan, created by or on behalf of a candidate or committee, that incorporates visual or audio media partially or wholly generated by AI.

Governance and Institutional Framework

The implementation and enforcement of Michigan's AI in Political Advertising Acts are primarily vested in existing state governmental bodies, leveraging their established authority over election integrity and campaign finance. The Michigan Secretary of State plays a pivotal role, particularly in overseeing the Michigan Campaign Finance Act, which is amended by these new laws. This includes the promulgation of rules concerning the size and placement of disclaimers required for AI-generated political advertisements, ensuring their clarity and conspicuousness across various media formats. The Secretary of State's office is instrumental in providing guidance and clarity to campaigns and individuals regarding compliance with the disclosure requirements, thereby facilitating a standardized application of the regulations across the state.

Furthermore, local election officials, in conjunction with the state apparatus, are integral to the broader oversight framework, particularly in identifying and addressing potential violations at the local level. The Michigan Attorney General is empowered to seek injunctive relief against individuals or entities violating the provisions related to AI disclosure or the distribution of materially deceptive media. This judicial recourse provides a critical mechanism for swift intervention to prevent the widespread dissemination of misleading content, especially during the sensitive period leading up to an election. Candidates who believe they have been harmed or are likely to be harmed by a violation also have standing to pursue injunctive relief, offering an additional layer of accountability and protection for those directly impacted by deceptive practices. This multi-tiered approach, involving administrative oversight, prosecutorial powers, and civil remedies, creates a robust institutional framework for upholding the integrity of political discourse in Michigan.

Key Focus Areas

The Michigan Public Acts 263-266 of 2023 establish two primary areas of focus: mandatory disclosure for AI-generated content in political advertisements and the prohibition of materially deceptive media (deepfakes) with electoral intent. Regarding disclosure, House Bill 5141 (Public Act 263 of 2023) mandates that any 'qualified political advertisement' that is created, published, or distributed, either wholly or substantially, with the use of artificial intelligence, must include a clear and conspicuous statement to that effect. The specific requirements for this disclaimer vary depending on the medium of the advertisement. For text or graphic communications, the statement must appear in letters at least as large as the majority of the text. In video communications that also contain audio, the disclaimer must appear for at least four seconds in easily readable letters and be spoken audibly and intelligibly for at least three seconds at the beginning or end. For audio-only communications, the disclaimer must be read at both the beginning and end of the message. Prerecorded telephone messages generated by AI also require a specific disclaimer. These detailed requirements aim to ensure that voters are explicitly aware when they are encountering AI-generated content.

The second key focus, addressed by House Bill 5144 (Public Act 265 of 2023), prohibits the distribution of 'materially deceptive media' within 90 days before an election if the distributor knows the media falsely represents an individual and intends to harm a candidate's reputation or electoral prospects, or to deceive voters into believing the depicted individual engaged in the false speech or conduct. This prohibition is critical for preventing last-minute disinformation campaigns that could significantly sway election outcomes. However, the act provides an important exception: the prohibition does not apply if the media includes a clear disclaimer indicating that it has been manipulated by technical means and depicts speech or conduct that did not occur. If the deceptive media is an edited version of existing content, it must also include a citation to the original source. This balance allows for creative or satirical uses of manipulated media while ensuring that genuinely deceptive content, especially deepfakes, is either clearly identified or prohibited, thereby protecting the public from malicious manipulation.

Implementation Framework

The implementation framework for Michigan's AI in Political Advertising Acts is designed to integrate new disclosure and prohibition requirements into existing campaign finance and election law structures. For disclosures related to AI-generated content, the acts specify the precise manner in which disclaimers must be presented across various media. For instance, in visual advertisements, the disclaimer must be clearly visible and readable, often requiring a minimum font size relative to other text or a specific duration for video and audio formats. This granular approach ensures that the disclosure is not easily overlooked or obscured, thereby maximizing its effectiveness in informing the public. The Michigan Secretary of State is empowered to issue further rules to clarify these requirements, particularly concerning the size and placement of identification or disclaimers, allowing for adaptability as new forms of political advertising emerge.

Regarding the prohibition on materially deceptive media, the acts establish a critical 90-day window before an election during which such content is strictly forbidden if intended to mislead voters. This timeframe is strategically chosen to prevent the last-minute dissemination of deepfakes, which could be difficult to debunk or counter before election day. The framework also outlines the conditions under which manipulated media may be distributed, specifically requiring prominent disclaimers and, in cases of edited existing media, a citation to the original source. This provision acknowledges legitimate uses of media manipulation while creating a clear standard for transparency. The acts also detail the legal avenues for enforcement, including civil actions for injunctive relief, which can be initiated by the Attorney General or by injured candidates, providing a rapid response mechanism to address violations and mitigate their potential impact on elections.

Monitoring and Evaluation

The Michigan Public Acts concerning AI in political advertising do not explicitly establish a new dedicated governmental body for continuous monitoring and evaluation. Instead, the framework relies on the existing oversight mechanisms inherent in Michigan's election and campaign finance laws, augmented by the new enforcement provisions. The Michigan Secretary of State, already responsible for administering campaign finance regulations, implicitly plays a role in monitoring compliance with the AI disclosure requirements through the review of campaign materials and responses to complaints. This existing administrative capacity is expected to absorb the new responsibilities, ensuring that AI-related violations are addressed within established regulatory processes.

Evaluation of the effectiveness of these acts will largely be derived from the enforcement actions undertaken and the outcomes of any legal challenges. The provision for injunctive relief, allowing the Attorney General or aggrieved candidates to seek court orders to stop the distribution of non-compliant or deceptive media, serves as a de facto monitoring mechanism. Each instance where such relief is sought and granted, or where penalties are imposed, will provide data points on the prevalence of AI misuse and the efficacy of the legal deterrents. While no formal, periodic review cycle is outlined within the acts themselves, the ongoing legislative process and public discourse surrounding AI regulation will likely prompt future assessments and potential amendments to ensure the laws remain relevant and effective in a rapidly evolving technological landscape. The fiscal impact of these bills, while indeterminate, suggests that the state anticipates some level of enforcement activity, which will contribute to the practical evaluation of the legislation.

Penalties, Liability, and Appeals

The Michigan Public Acts introduce a tiered system of penalties and mechanisms for liability and appeals to deter the misuse of AI in political advertising. For violations of the AI disclosure requirements under House Bill 5141 (Public Act 263 of 2023), a first offense is classified as a state civil infraction, potentially incurring a civil fine of up to $250. Subsequent violations escalate to a state civil infraction with a fine of up to $1,000 for each offense. If a committee is found in violation, it faces a misdemeanor charge, punishable by up to 93 days imprisonment or a fine of up to $1,000, or both. Furthermore, House Bill 5142, a companion bill, amends sentencing guidelines to classify a third or subsequent offense for failing to disclose AI use as a Class G felony against the public trust, carrying a potential imprisonment of up to two years. This progressive penalty structure aims to provide a strong deterrent against repeated non-compliance.

For the distribution of 'materially deceptive media' under House Bill 5144 (Public Act 265 of 2023), the penalties are more severe due to the direct intent to mislead voters. While the initial offense may be a misdemeanor, House Bill 5145 (Public Act 266 of 2023) amends the sentencing guidelines to make a subsequent violation within five years a Class E felony against the public trust, punishable by a statutory maximum term of imprisonment of five years. This significant criminal penalty underscores the gravity with which Michigan views the intentional dissemination of deepfakes to influence elections. Beyond criminal and civil penalties, the acts also provide for civil remedies. The Attorney General or any candidate for office who has been or is likely to be injured by a violation can apply to the circuit court for injunctive relief. This allows for immediate legal action to halt the distribution of prohibited content, preventing further harm to the electoral process or individual candidates. The availability of injunctive relief offers a crucial mechanism for rapid response, complementing the punitive measures by enabling proactive intervention.

Relationship to Other Instruments

The Michigan Public Acts 263-266 of 2023 are not standalone pieces of legislation but are intricately woven into the existing legal fabric of the state. They primarily function as amendments to three foundational Michigan statutes: the Michigan Campaign Finance Act (1976 PA 388), the Michigan Election Law (1954 PA 116), and the Code of Criminal Procedure (1927 PA 175). House Bills 5141 and 5143 specifically amend the Michigan Campaign Finance Act, integrating the new definitions of artificial intelligence and the disclosure requirements for AI-generated political advertisements directly into the state's established campaign finance regulations. This ensures that the new rules are administered and enforced within a familiar legal framework, leveraging existing institutional knowledge and processes.

House Bill 5144 amends the Michigan Election Law, introducing the prohibition against materially deceptive media and outlining the conditions for its permissible use with disclaimers. This integration into election law reinforces the state's commitment to maintaining the purity and integrity of elections. Finally, House Bills 5142 and 5145 make complementary changes to the Code of Criminal Procedure, establishing the sentencing guidelines and penalties for violations related to both AI disclosure and the distribution of deepfakes. These tie-barred provisions ensure that the criminal justice system is equipped to address offenses under the new acts, providing clear legal consequences. The tie-barring mechanism, explicitly stating that certain bills cannot take effect without the enactment of others, highlights the interconnected and interdependent nature of this legislative package, ensuring a cohesive and comprehensive approach to regulating AI in political contexts.

International Alignment

As state-level legislation within the United States, the Michigan Public Acts on AI in political advertising do not directly align with or implement specific international treaties, standards, or frameworks. The legislative process and scope are inherently confined to the jurisdiction of Michigan. However, these acts contribute to a broader global discourse and trend towards regulating artificial intelligence, particularly concerning its ethical implications and potential for misuse in democratic processes. Many nations and international bodies are grappling with similar challenges regarding deepfakes, disinformation, and the impact of AI on elections.

By establishing clear disclosure requirements and prohibitions, Michigan's legislation indirectly resonates with principles advocated by international organizations and expert groups that call for greater transparency, accountability, and the prevention of manipulative practices in the digital information environment. While not a direct implementation of international law, the Michigan acts reflect a shared concern across democracies about protecting electoral integrity in the age of advanced AI. They serve as an example of how sub-national jurisdictions are proactively responding to these emerging threats, potentially informing future national or even international policy discussions and best practices in the regulation of AI in politically sensitive contexts. The development of such laws at the state level demonstrates a localized effort to address issues that have global implications for democratic governance.

Implementation Timeline

Compliance Checklist

Sources and References