United States public health agencies are moving to formally test generative AI models from OpenAI, Anthropic, and other developers through structured pilots, federal safety evaluations, and coordinated public-private partnerships designed to evaluate the technology’s capabilities, risks, and practical value across real public health applications.
Central to these efforts is the Public Health Use Case and Learning Scaling Engine, known as PULSE, a program established to support generative AI trials across public health departments nationwide. The Coalition for Health AI is coordinating PULSE in partnership with enterprise AI products from OpenAI, Anthropic, and Accenture, directing resources toward public health practitioners.
Pilots are planned for ten state, local, tribal, or territorial public health jurisdictions, with the Coalition for Health AI leadership council responsible for selecting participating jurisdictions and assigning practitioners to use-case communities. PULSE pilots are scheduled to begin in autumn 2026, with implementation playbooks for public health agencies expected to follow in 2027.
The use cases targeted within PULSE reflect a broad range of public health priorities. Biosurveillance and drug-wave prediction have been identified as primary applications, alongside social determinants of health mapping to inform interventions. Operations, efficiency, and community-feedback analysis are included to support departmental optimization.
Public communications and multilingual translation represent another focus area, aimed at improving outreach to diverse populations. An automated clinical-data retrieval and FHIR query engine is also planned to streamline data access within public health decision-making workflows.
Federal safety testing of leading AI models is proceeding in parallel. Both OpenAI and Anthropic have entered agreements with the US Artificial Intelligence Safety Institute, housed within the National Institute of Standards and Technology, to provide access to major models before and after public release for research, testing, and evaluation.
The US AI Safety Institute has also established a collaboration with its UK counterpart to conduct joint safety evaluations and provide feedback on frontier models, extending the scope of international oversight.
The Centers for Disease Control and Prevention has issued guidance specific to generative AI use in public health, requiring human oversight, accuracy checks, and review of AI outputs for hallucinations or misleading content. The CDC additionally mandates verification of citations and sources in AI-generated material before any use in public health contexts.
Beyond public health, the Food and Drug Administration has announced a proposed pilot program to evaluate how AI-enabled technologies can improve efficiency, speed, and decision quality in early-phase clinical trials, signaling broader federal interest in assessing AI across health-related regulatory functions.
OpenAI and Anthropic have each donated 10 enterprise licenses to support PULSE, making access available to up to 2,000 public health practitioners across participating jurisdictions.
Taken together, these initiatives reflect a coordinated federal approach to evaluating generative AI under real operational conditions, with frameworks emphasizing safety, equity, transparency, and rigorous evaluation as prerequisites for trustworthy adoption across government public health operations.
