By McKaia Dykema

As artificial intelligence (AI) tools become more common, Minnesota cities are exploring how these technologies can improve efficiency, customer service, and daily operations. From drafting meeting minutes more quickly to generating city-branded graphics and designs, AI offers clear benefits.

For local governments, however, these opportunities come with important legal obligations, particularly under the Minnesotan Government Data Practices Act (MGDPA). Understanding how the MGDPA applies to AI use is essential for cities seeking to leverage new technologies while maintaining public trust and compliance.

Government data defined

Under the MGDPA, “government data” includes all data collected, created, received, maintained, or disseminated by a government entity, regardless of format. This broad definition may include data entered into AI tools, AI-generated outputs, and other metadata.

Government data encompasses a wide range of formats, including computerized files, emails, photographs, charts, maps, audio recordings, video recordings, and even handwritten notes or working documents.

As a result, the MGDPA’s data classifications, protections, and management requirements may apply to information used in — or produced by — AI tools and systems.

Is AI data public or nonpublic?

The MGDPA establishes a presumption that government data is public unless otherwise classified by law. Data that is “not public” includes information classified as confidential, private, nonpublic, or protected nonpublic.

Whether AI-related data is public depends on the classification of the underlying data. In other words, cities must evaluate, on a case-by-case basis, whether data used in or produced by AI falls into a protected or public category.

Proper classification of both data entered into AI tools and data generated by those tools is essential to ensure compliance with MGPDA requirements for access, security, and protection.

Entering data into AI tools

A key concern when using AI tools is what happens to government data once it is entered into third-party systems. Many systems retain user inputs on their servers, sometimes to improve their models. This creates a risk that cities may lose control over government data once it is submitted.

For this reason, the League of Minnesota Cities recommends cities understand the classification of data before entering it into an AI system and limit inputs to public data. Entering nonpublic data into AI tools could constitute a data breach. This includes but is not limited to Social Security numbers, personnel records, and attorney-client privileged information.

Cities should also avoid entering operationally sensitive data for security reasons, IT system details, security metadata, surveillance footage, and technical specifications.

If a city intends to use AI tools with nonpublic data, it should first consult with its city attorney and ensure appropriate contractual, technical, and legal safeguards are in place to meet obligations under the MGDPA.

Impact on data requests

The MGDPA requires cities to respond to requests for access to government data. When AI tools are used, responsive data may include AI-generated records, documents, system data (such as input history), or outputs, if they fall within the scope of a request.

Cities should assume AI-related data may be subject to disclosure and manage it accordingly. When questions arise about data classification or whether certain AI-related records are responsive to a request, cities should consult their city attorney.

Practical considerations

Cities using or considering AI tools should take proactive steps to align their practices with MGDPA requirements.

These steps may include:

• Developing internal policies governing AI use, including approved tools and prohibited data types for inputting into systems. Access the League’s computer-use model policy that includes references to AI, as well as sample AI policy language, at lmc.org/AIconsiderations.
• Training staff on data classification and the risks associated with AI platforms.
• Reviewing vendor contracts, when applicable, with your city attorney to ensure data ownership, storage, security, and MGDPA compliance terms are clear and legally enforceable.

Artificial intelligence presents meaningful opportunities for Minnesota cities, but the use of this evolving technology does not change existing legal responsibilities. Cities that approach AI thoughtfully — and ground their practices in MGDPA compliance — will be best positioned to benefit from these tools while maintaining transparency, protecting privacy, and preserving public trust.

McKaia Dykema is a staff attorney with the League of Minnesota Cities. Contact: [email protected] or (651) 281-1261.