The State of AI in Public Health: New Data from the 2025 ASTHO Profile
April 28, 2026
The Association of State and Territorial Health Officials’ (ASTHO) Profile of State and Territorial Public Health (ASTHO Profile) has served as the authoritative source for data on state and territorial public health agency activities, structure, and workforce since 2007. In anticipation of the 2025 ASTHO Profile release later this year, we are pleased to provide an early look at how state and territorial health agencies are navigating the rise of artificial intelligence (AI). Based on responses from 41 state and territorial health offices (n=32 to 44, depending on the question), the data reveals a public health landscape in transition, with a significant divide in how agencies approach and utilize these new tools.
AI Policy and Oversight
A majority of agencies have established some form of policy framework for AI, though the origins of these policies vary.
- Policy Adoption: 52% of agencies operate under a statewide policy, while 11% have developed an agency-specific policy.
- In Development: 9% of agencies currently have a policy in development, while 25% have not yet adopted one.
- Policy Focus: Among the 28 agencies with policies, Data Governance, Privacy, and Security is the most common topic addressed (81%).
- Other Topics: Policies also frequently cover: Evaluation and Accountability (59%) and Use Case Identification (53%), though only 34% address Leadership and Workforce Readiness.
Current AI Usage and Tools
Agencies generally use AI for administrative tasks rather than specialized public health functions.
- Primary Use Cases: The most common applications are Administrative and Operational Efficiency (30%) and Content/Report Creation (30%).
- Specialized Tasks: Only 14% of agencies report using AI for Disease Surveillance, Anomaly Detection, or Emergency Response.
- Software Types: 23% of agencies utilize consumer generative AI tools, while 32% use enterprise AI environments.
- Non-Users: Approximately 34% of agencies report they are not using AI systems or software at all. How AI is Being Deploy
How AI is Being Deployed
Public health currently uses AI for broad administrative support and highly specialized technical initiatives. At the federal level, CDC has successfully used enterprise AI solutions for a variety of use cases including analyzing unstructured grant reports, saving over 5,500 labor hours. Similarly, a small number of states, such as Minnesota and California, leverage AI to conduct real-time disease surveillance that can identify outbreaks weeks ahead of traditional systems. However, this level of integration is rare at the state and territorial level. A significant gap exists between those using consumer-grade products and those with enterprise systems. While 23% of agencies use consumer tools for non-sensitive work, 32% have invested in enterprise environments that can safely process electronic health records for surveillance and outbreak response.
The Path Toward Integration
The 2025 ASTHO Profile highlights a field in flux, where policy creation has outpaced practical implementation. While a third of agencies don’t report using AI at all, the majority are experimenting with administrative use cases while facing significant barriers in workforce skills and resource constraints. Bridging this gap will require a shift from individual best efforts toward shared infrastructure and expanded AI literacy across the public health workforce.
ASTHO's Support for Data Modernization and AI
ASTHO actively supports jurisdictions through its broader Data Modernization Initiative, viewing AI as a critical component of a modernized, response-ready data ecosystem. By transitioning from manual to automated processes, AI helps agencies handle the increasing volume and velocity of public health data more efficiently. At a time with real workforce constraints, solutions that streamline tasks and save time are important opportunities. To assist jurisdictions in their AI journey, ASTHO provides direct technical assistance to offer hands-on implementation support. Agencies can also leverage the Informatics Directors Peer Network to share promising practices and navigate the ethical use of these tools.
2025 ASTHO Profile Survey Question and Answers
| Answer | Number of Respondents (n) | Percentage of Respondents (%) |
|---|---|---|
| Yes, as part of a statewide policy | 23 | 52% |
| Yes, agency-specific policy | 5 | 11% |
| No, but there is a policy in development | 4 | 9% |
| No, we have not adopted an AI policy | 11 | 25% |
| Unsure | 1 | 2% |
| Other, please specify | 2 | 5% |
| Answer | Number of Respondents (n) | Percentage of Respondents (%) |
|---|---|---|
| Data governance, privacy, and security | 26 | 81% |
| Bias | 11 | 34% |
| Leadership and workforce readiness | 11 | 34% |
| Use case identification and testing | 17 | 53% |
| Evaluation and accountability | 19 | 59% |
| Transparency and "explainability" | 16 | 50% |
| Technical readiness | 11 | 34% |
| Other | 4 | 13% |
| Answer | Number of Respondents (n) | Percentage of Respondents (%) |
|---|---|---|
| Administrative and operational efficiency | 13 | 30% |
| Content and report creation | 13 | 30% |
| Chatbots and conversational agents | 9 | 20% |
| Other | 8 | 18% |
| Generating or correcting code | 8 | 18% |
| Communications campaigns and community engagement | 7 | 16% |
| Anomaly detection, disease surveillance, or predictive modeling (combined) | 6 | 14% |
| Human resources support | 4 | 9% |
| We are not using AI | 14 | 32% |
| Answer | Number of Respondents (n) | Percentage of Respondents (%) |
|---|---|---|
| Consumer generative AI tools (e.g., OpenAI’s ChatGPT, Google’s Gemini, Anthropic’s Claude) | 10 | 23% |
| Enterprise AI tools (including vendor-developed or developed in-house) | 16 | 32% |
| We are not using AI systems/software within our public health agency | 15 | 34% |
| Other (e.g. CoPilot included in Microsoft 365 subscription) | 6 | 14% |
| Answer | Number of Respondents (n) | Percentage of Respondents (%) |
|---|---|---|
| Lack of established guidance around public health use of AI | 28 | 64% |
| Lack of workforce skills or knowledge | 24 | 55% |
| Concerns about AI-based data accuracy | 20 | 45% |
| Lack of resources (e.g., funding, staff, partners) | 17 | 39% |
| Legal or policy barriers | 17 | 39% |
| Concerns related to fairness and representativeness of data | 13 | 30% |
| Lack of evaluation data about tested use cases | 12 | 27% |
| Other initiatives are more important | 11 | 25% |
| Technical challenges (platform integration or interoperability) | 9 | 20% |
| Other | 7 | 16% |
| Lack of coordination within the health department | 6 | 14% |
| Lack of public trust in AI technology | 6 | 14% |
| Lack of leadership or staff support/buy-in | 4 | 9% |
| We have not encountered any challenges | 4 | 9% |
Key ASTHO AI and Data Resources
- Data Modernization Primer and Tactical Guides: Strategic steps for planning and sustaining modernized data systems.
- How to Modernize Data Infrastructure Toolkit: A structured approach for health leaders to build resilient technology systems.
- ASTHO AI Legislative Tracking: A tool for monitoring state and territorial legislation regarding government AI use.
- AI in Public Health Brief: Insights into how informatics leaders are currently leveraging AI for anomaly detection and content generation.
Reviewed by Tabatha Offutt-Powell, Vice President, Public Health Data Modernization and Informatics.
This resource was developed with support from the Robert Wood Johnson Foundation and the Centers for Disease Control and Prevention. Its contents are solely the responsibilities of the authors and do not necessarily represent the official views of the Robert Wood Johnson Foundation, the Centers for Disease Control and Prevention, nor the Department of Health and Human Services.