The Healthy Technology Act of 2025 authorizes FDA-approved artificial intelligence and machine learning systems to prescribe medication in states where such technology is legally permitted.
David Schweikert
Representative
AZ-1
The Healthy Technology Act of 2025 amends the Federal Food, Drug, and Cosmetic Act to authorize artificial intelligence and machine learning systems to prescribe medication. To qualify, these technologies must receive FDA approval and comply with specific state-level regulations.
The Healthy Technology Act of 2025 seeks to fundamentally change who—or what—can write your next prescription by amending Section 503(b) of the Federal Food, Drug, and Cosmetic Act. The bill proposes recognizing artificial intelligence and machine learning (ML) systems as 'licensed practitioners' with the legal authority to prescribe medication. To start issuing scripts, an AI system would need to meet two criteria: the state must pass a law allowing AI to prescribe that specific drug, and the software must be cleared or approved by the FDA through existing medical device pathways like the 510(k) or 515 processes.
This shift moves AI from a helpful tool used by doctors to a standalone decision-maker. In practice, this could mean a patient interacting with a software interface that analyzes their symptoms and automatically sends a prescription to the pharmacy without a human doctor ever reviewing the case. For a busy parent or a shift worker in a rural area with no local clinic, this might look like a win for convenience and speed. However, because the bill relies on existing FDA medical device pathways (Sections 510, 513, 515, or 564), it raises questions about whether these processes—originally designed for hardware like pacemakers or diagnostic software—are robust enough to evaluate the complex, evolving logic of a machine that is now authorized to act as a primary prescriber.
The most significant change is the potential removal of human oversight in the high-stakes moment of prescribing medicine. While a human doctor considers a patient's nuanced history and physical cues, an AI relies strictly on data inputs. If an algorithm makes an error—perhaps missing a rare drug interaction or misinterpreting a symptom—the bill is silent on who bears the legal or professional responsibility. For healthcare professionals, this represents a major shift in the definition of medical practice, potentially moving the 'licensed practitioner' status away from individuals with years of clinical training and toward software developed by tech companies.
Even if this passes federally, the rollout would be a patchwork across the country. Since the bill requires specific state-level permission for each drug an AI prescribes, we could see a reality where an AI can prescribe an antibiotic in one state but is legally barred from doing so across the border. This creates a complex regulatory environment for both patients and the tech companies building these systems. While the goal may be to increase efficiency and lower costs in a strained healthcare system, the trade-off is a new frontier of patient safety where the 'doctor' is a line of code, and the traditional safety net of human judgment is no longer a requirement.