Considerations for Generative AI in Public Health

What is Generative AI?

GenAI refers to a class of artificial intelligence (AI) that can create new or synthetic content, such as text, images, audio, or code. This informational resource focuses primarily on text-based applications most relevant to public health operations. These tools respond to prompts and are well-suited for open-ended tasks like writing, summarizing, translation, and analysis. Conversational GenAI products powered by large language models (LLMs) are already being used across public health agencies to improve communication, streamline operations, and support early-stage research and planning. While GenAI encompasses text, images, audio, and code generation, this information focuses primarily on text-based applications most relevant to public health operations.

When GenAI Works Best

GenAI can be used for well-scoped, content-driven tasks like drafting, rewriting, summarizing, or synthesizing text, when the purpose and audience are clear and the output will be reviewed before use1. GenAI should be treated as a drafting and synthesis aid, not an authority2. It is not a substitute for subject-matter expertise, validated analytic methods, or secure systems for handling sensitive data12. Human oversight and clear expectations are essential because generative models can confidently produce incorrect or fabricated content, and higher-risk uses require defined oversight processes123.

CDC Effective Use Principles

The following principles support the effective use of GenAI tools in public health and are adapted from CDC's internal guidance. They align with the NIST AI Risk Management Framework, which emphasizes transparency, accountability, and oversight. While these principles guide CDC's approach, they are not requirements for STLT agencies; rather, they can serve as a model for agencies seeking to adopt GenAI.

• Mission Alignment: Tie use to a clear public health goal such as improving efficiency, informing decisions, or enhancing communication.
• Human Oversight: Always review GenAI outputs before use. A person should be accountable for the final product.

• Safety and Reliability: Review GenAI outputs for accuracy, completeness, and hallucinations or misleading content. Confirm citations are valid and appropriately sourced.
• Scientific Integrity: Ensure outputs are accurate, maintain scientific rigor, are free from fabrication or plagiarism, and note limits and uncertainties.
• Disclosure and Transparency: Disclose when GenAI is used in public-facing or significant internal products. Use disclaimers or attribution statements where appropriate.
• Security and Privacy: Follow agency's data security policy; Non-public data may be used only with agency-approved enterprise environments with explicit authorization and security/privacy review.

Foundational Steps for STLT GenAI Adoption

CDC's experience with GenAI implementation4 shows that public health agencies can benefit from careful planning, cross-functional coordination, and secure data practices. These steps reflect CDC learnings on data governance and early adoption patterns among STLT agencies. The actions below are considerations for agencies preparing to adopt GenAI in a mission-aligned way.

1. Engage Leadership and Key Stakeholders
Secure leadership support and involve stakeholders from IT, legal, privacy, communications, and program areas. Early alignment helps ensure policies, resources, and expectations are clearly defined.

2. Establish Access to Secure, Compliant Tools
Identify GenAI tools that meet your agency's data protection and security requirements. Consider starting with FedRAMP-authorized platforms or tools provided through HHS or state-level procurement.

3. Establish or Expand an Agency-Wide GenAI Policy and Use Guidelines
Create concise, practical guidance for leadership and staff. Define appropriate use cases, prohibited uses, and expectations for human oversight and review. For examples refer to the: CDC Effective Use Principles and Kansas Health Institute's Developing Artificial Intelligence (AI) Policies for Public Health Organizations.

4. Build Staff Capacity Through Training and Peer Learning
Offer training on GenAI fundamentals, risk management for potential misuses, and related AI capabilities such as deep research tools (see Considerations for Agentic Research in Public Health). Encourage peer-to-peer learning through user groups, office hours, or communities of practice.

5. Monitor Use and Share Outcomes
Track how GenAI is being used and assess its impact. Share successful use cases internally and with partners to support continuous learning and build trust.

STLT Best Practices in Motion: Real-World GenAI Use Across STLT Public Health Agencies

Responses to CDC's 2025 Epidemiology and Laboratory Capacity (ELC) Health Information Systems Survey show that many STLT public health agencies that have adopted AI are already using GenAI tools in day-to-day work. Agencies report using these tools to:

• Draft and edit documents and reports.
• Summarize meeting recordings and notes.
• Generate or correct code for data and IT tasks.
• Create and refine content for internal and external communication.
• Pilot GenAI-enabled chatbots and digital assistants to handle common questions.

While governance and Return on Investment (ROI) methods are still evolving, agencies consistently report time savings, reduced administrative burden, and improved documentation quality, allowing staff to reallocate effort to higher-value public health activities.

The following examples illustrate how STLT agencies are putting these capabilities into practice. Each reflects practical, low-risk GenAI applications that align with the learnings in this document, demonstrating how agencies can begin with clear, contained workflows and expand over time.

Rapid GenAI-Supported Synthesis for Strategic Planning (State Health & Human Services)

One State Department of Health and Human Services piloted GenAI to process transcripts from its data modernization strategic planning sessions, where leadership and program staff had logged more than 60 hours of meetings. Staff reported that GenAI could summarize transcripts, identify cross-cutting themes, and surface recurring needs in roughly 1.5 hours, compared with an estimated seven hours of manual review per session. Program staff still performed human review to extract key quotes, clarify low-quality audio, and confirm themes, so GenAI was framed as a drafting and synthesis aid rather than a final analytic authority. This example shows how STLT agencies can use GenAI to handle large volumes of qualitative input while maintaining human ownership of interpretation and narrative framing.

Source: The Rapid GenAI-Supported Synthesis for Strategic Planning Case Study is based on learnings reported to CDC by STLT partners as part of the CDC STLT Data Connection (July 2025).

Improving Privacy Workflows Through GenAI-Assisted Redaction and Document Cleanup (Multiple STLTs)

Several STLT agencies reported early but active experiments with GenAI tools to streamline text-heavy tasks such as cleaning up memos and reports, redacting sensitive information, and preparing polished drafts for internal use. These agencies emphasized keeping humans involved in all privacy-related steps but noted that GenAI reduced time spent on routine review and improved overall clarity of documentation. Early pilots show how GenAI can strengthen daily operations while supporting policy-aligned use.

Source: The Improving Privacy Workflows Through GenAI-Assisted Redaction and Document Cleanup Case Study is based on learnings reported to CDC by STLT partners as part of the CDC STLT Data Connection on (July 2025).

Streamlining Health Communications with GenAI Tools (Local Health Departments)

Local health departments (LHDs) increasingly report using GenAI tools to draft plain-language advisories, multilingual outreach messages, and social media content related to respiratory season, heat events, and other public health concerns. These drafts are consistently routed through established communication and leadership review processes, but GenAI significantly shortens initial drafting time and provides useful variations for different audiences. National survey data from the National Association of County and City Health Officials (NACCHO) indicate that communications drafting is already the most common use of GenAI among LHDs currently adopting AI tools.

Source: NACCHO. 2024 Public Health Informatics Profile Report.

Using GenAI to Improve Case Documentation and Intake in County Human Services

County human services agencies use GenAI-powered tools to support case documentation, draft summaries from structured data, reduce repetitive entry, and streamline referral preparation. Staff remain the "author of record," but GenAI helps reduce administrative bottlenecks and frees time for direct client engagement. These early implementations show how GenAI can be safely integrated into back-office workflows when paired with human oversight and clear governance.

Source: National Association of Counties (NACo). AI County Compass. December 2024.

CDC’s Approach: Enablers of Successful GenAI Adoption

Public health agencies are dedicated to improving the lives of the American people. To support this mission, CDC adopted GenAI in 2023 to empower staff, accelerate public health work, and expand capacity across programs. Rather than taking a one-size-fits-all approach, CDC implemented a phased and secure rollout that emphasized effective use and measurable impact, including:

• Estimated upwards of 41,000 staff hours saved annually (based on FY25 usage) through CDC's internal GenAI tool, primarily from tasks involving content creation and coding based upon an internal study focused on usage rates and categories of usage5.

• 103 AI use cases reported in HHS's 2025 AI Inventory, including automated document review for outbreak response, literature synthesis for emerging threats, and public-facing tools like data assistants that expand community access to health information.

CDC's success with GenAI was built on deliberate planning, strong leadership, and integration with existing systems, not just tool access. These enablers allowed the agency to scale GenAI use across programs while maintaining trust, transparency, and oversight. STLT public health agencies may find these insights helpful as they design their own approach to GenAI adoption.

• Early leadership engagement: Executive buy-in and active sponsorship enabled rapid resource allocation, policy development, and cultural change management.
• Phased rollout approach: CDC piloted tools with select use cases, established enterprise agreements with FedRAMP-authorized vendors and implemented data governance controls before general availability. This allowed testing, gathering feedback, refining functionality, and building confidence before broad deployment.

• Balance of enablement and guardrails: CDC emphasized enabling access while establishing clear principles. Rather than restrictive policies, CDC focused on secure systems, human oversight, and transparency requirements that allowed innovation within appropriate boundaries.

• Integration with existing governance: CDC embedded AI governance within existing IT, data, and scientific review processes rather than creating parallel structures. This reduced administrative burden and accelerated adoption.

• Transparent communication: Public disclosure of AI use, both at agency and product levels, built trust with partners and the public while normalizing effective AI adoption.

• Community-driven learning: AI Community of Practice and peer learning networks enabled grassroots knowledge sharing, use case discovery, and rapid problem-solving across the agency.

Resources

CDC and Federal Resources

These resources can help public health agencies explore, implement, and govern GenAI tools.

HHS AI Use Case Inventory Examples of how GenAI is being used across HHS programs, including CDC.

FedRAMP Marketplace A list of cloud service providers authorized for federal use, including platforms.

National Institute of Standards and Technology (NIST) AI Risk Management Framework A framework for managing AI risks, including governance, transparency, and safety.

GSA OneGov Buy AI Provides a list of OneGov agreements and GSA contracting vehicles to acquire AI solutions. As of March 11, 2026, CDC has access to 2 enterprise GenAI tools through HHS and OneGov.

NIST AI Risk Management Framework Crosswalk

As your agency develops its own guidance, governance, and policies for GenAI, the NIST AI Risk Management Framework can help align those efforts with a nationally recognized standard. This crosswalk shows how recommended STLT practices map to the NIST AI RMF's four core functions: govern, map, measure, and manage. This can be used it to identify where your current or planned activities align with the framework and to support a more structured approach to managing GenAI-related risks.