Artificial Intelligence tools show promise for improving health care. They can help predict health trajectories, recommend treatments, and automate administrative tasks. Challenges associated with these tools include: Transparency: If a medical provider doesn't know how a tool works, it could reduce trust in the tool. Bias: Limitations and bias in data can reduce the safety and effectiveness of AI tools. Data: Obtaining the high-quality data needed to create effective AI tools can be difficult. We offer policy options—such as improving data access, establishing best practices, and more—to address these and other challenges we found.