By: Natalie Johnson
Texas Senate Bill 1188, effective September 1, 2025, requires a healthcare practitioner who uses artificial intelligence (AI) for diagnostic purposes to disclose that use to patients. Walk into a doctor’s office in New York today, and an automated tool may be listening to the conversation or reading medical images. New York has no equivalent disclosure rule, so whether a patient hears about that tool depends on the state where the visit takes place. That gap between state lines is a growing challenge for healthcare leaders who want to move quickly with new technology while keeping patient trust at the center of care.
Hospitals and local clinics are adopting automated tools to help doctors stay on top of their day. Automated charting tools listen to patient visits and turn spoken conversations into draft clinical notes, while other programs review scans and flag findings before a radiologist confirms them. Chris Calitz points out that these tools are already in use in major health systems, and he counts that as good news. As he explains, “AI tools are widely used in healthcare settings across the United States, including the state of New York.” What is missing is disclosure.
Two States, Two Different Playbooks
The market assumes a process built for one location will satisfy the rules in another, but without a single national framework, individual states are building their own. Texas has taken a direct approach by placing clear responsibilities on the physician practice using these systems. New York has taken a different path, with early legislation focused on areas such as insurance decisions and the companies developing large AI models. None of it yet requires a clinician to tell a patient that AI helped with their care.
The result is a patchwork that gives healthcare organizations different requirements depending on where they operate. In the absence of a federal standard, Calitz says, “You see this kind of jigsaw puzzle of different laws across the country.” For healthcare executives, adoption is settled: these tools are already in clinical workflows. The open question is whether every tool in use has a clear owner, a defined purpose, and a process for checking how it performs.
Transparency in Clinical Workflows
When no one mentions the technology, the patient loses an important opportunity to understand how their care is being supported. Informed conversations depend on patients knowing enough to ask reasonable questions about the tools involved in their care. A simple explanation can make the technology part of the conversation rather than something patients discover later.
That matters because trust in healthcare depends on more than the final diagnosis. Patients need to understand how decisions are made and who is responsible for them. If someone discovers after the fact that an automated system helped produce a clinical note or support a diagnosis, the immediate question may be why that information was never mentioned. Calitz describes the underlying issue as “information asymmetry,” where one side knows that AI is involved and the other does not. The solution does not require a long technical explanation. It can be as simple as telling the patient what the tool does, what the physician reviews, and where human judgment remains responsible for the final decision.
Testing AI Tools is Critical
The way states assign responsibility also changes how medical groups evaluate new software. When the physician practice remains responsible for the technology it puts into a clinical workflow, the practice has a direct reason to understand how that system performs before relying on it. That responsibility should extend beyond reading a vendor’s sales material. A clinical organization can establish its own process for checking accuracy, reviewing outputs, documenting problems, and deciding when to change or remove a tool.
The goal is to scale up AI faster, under controls the organization can show. Calitz frames the operational reality: “The obligation to test the accuracy and reliability of the AI tools used in healthcare settings is owned by the physician practices, not the vendors.” That ownership gives clinical leaders a practical way to manage new technology. Instead of treating governance as a separate exercise that happens after deployment, the organization can build Audit-Based Controls into the workflow from the start: oversight that operators, clinical boards, and regulators can see and trust in real time.
Setting Clinical Standards
Medical leaders do not need to wait for a state law before they start being open with patients. A practice can train its staff to answer basic questions about how automated tools support clinical work, who reviews the output, and what happens when the system gets something wrong. The strongest approach is to explain the technology before the patient has to ask. A short explanation removes uncertainty and allows the physician to reinforce the role of human judgment. It also gives the organization a consistent answer when patients, staff, or board members ask how a new tool is being used.
What moves a practice to act is the peer organization down the road that can already answer these questions. Until lawmakers establish a clear national standard, healthcare executives can set a practical internal standard of their own and test it next quarter: give every clinical AI tool an owner, a defined use, a process for checking its output, and a plain explanation for patients. The best technology policy is one the patient can understand.
Follow Chris Calitz on LinkedIn or visit Amplify Impact Consulting at www.amplify-impact.ai.
Disclaimer: This article is for general informational purposes only and does not constitute legal advice. Laws and regulations change and vary by jurisdiction, and the application of any law depends on the specific facts involved. Readers should consult a qualified attorney for advice regarding their own situation.
