Why healthcare security leaders are paying attention
HIPAA Journal reports that healthcare security leaders are increasingly concerned about employees sharing patient information with AI tools. Their concerns include loss of patient privacy and potential HIPAA violations. The report describes a growing concern, not a specific breach at a medical or dental practice.
For a small practice, the issue can come up during routine work. Imagine a staff member asking an AI tool to make an appointment reminder sound friendlier. They can ask for a generic template without including patient information. But if they paste an actual message to save time, they may include details the tool doesn't need. The same distinction applies to asking for a blank form versus uploading a completed one.
Make the rule easy to use during a busy day
A policy is only useful if someone at the front desk knows what it means while answering calls and checking in patients. Start with a rule staff can apply without having to make a privacy judgment on the spot: Don't enter patient information into an AI tool unless the practice has explicitly approved both the tool and that specific use. If you're not sure, stop and ask the person responsible for privacy at your practice.
That rule doesn't have to put routine writing tasks on hold. Staff can work from blank templates or made-up examples rather than starting with a real patient message and trying to decide which details to remove.
Four steps for practice managers
- Find out where AI is already part of the work. Ask staff what they use it for, including drafting messages, summarizing text or reworking forms. Ask about tools they open separately and features built into software they already use. Make this a fact-finding conversation so staff feel comfortable telling you what's actually happening.
- Write down what you approve, not just what you prohibit. For each approved tool, list the tasks staff may use it for and what information they may enter. If a tool is approved for generic office writing, say plainly that this does not extend to patient messages or records. Give staff one place to check the current rules and one person to ask about a new use.
- Practice with an example from your office. At a staff meeting, compare asking AI to draft a generic follow-up reminder with pasting a real patient's message and asking for a rewrite. Show employees how to get the result they need using a blank template. Include uploads as well as text pasted into a chat so the rule applies to both.
- Give staff a way to report a mistake promptly. Tell employees whom to contact if they think they've shared patient information with an AI tool. Ask them to note the tool they used, when it happened and what kind of information was involved, without copying the patient details into another message. Have the person responsible for privacy review what happened and decide what follow-up is needed.
