← Insights

What Are the Benefits and Risks of AI in Telehealth?

AI already helps doctors save time and check more options. Here is what it does well, how your information is kept safe, and what to ask.

7 min read

By Bella Fan, Marketing Specialist

Reviewed by Bella Fan, Marketing Specialist ·

What Are the Benefits and Risks of AI in Telehealth?

AI already helps with telehealth visits. It writes the notes, so your doctor can look at you. It points out things the doctor might want to check. It helps people who speak another language. And it helps people stay in touch between visits. The best results come when a real doctor checks the AI's work and the company follows strict privacy rules.

What are the benefits of AI in telehealth?

AI saves doctors time, and it helps them think through more options. Here is what we know so far.

Less typing, more looking at you

More than one quarter of US doctors now use AI note-takers during visits, according to Medscape. The tool listens and writes a draft note. The doctor reads the draft before it goes into your file.

Mara Gordon, a family doctor who answers reader questions for NPR, says it helps her a lot. She used to type and talk at the same time. Now she can look at her patient while they talk.

A longer checklist

A 2026 review looked at AI tools that help doctors during video and phone visits. In one set of 102,059 visits, the AI's list of possible diagnoses included the doctor's final answer 84.2% of the time. The reviewers said that check was not fully independent, so treat it as a good sign, not proof. It also does not tell us how often the AI ranked the right answer first. Still, a longer list helps a doctor avoid missing something.

Better advice in urgent care

In the same review, AI-suggested urgent care advice was rated best more often than the doctors' final advice: 77.1% compared with 67.1%. The studies are small, so bigger studies need to check this.

BenefitWhere you see itHow sure we are
Faster notesDoctor visitsMany doctors use it and say it saves time
A longer list of possibilitiesVideo and phone visitsA good sign, from six studies
Better advice in urgent careUrgent care visitsPromising, but needs bigger studies

Help for people who speak another language

The US health department, HHS, gives an example in its guidance on phone and audio visits. A doctor can use an app that translates what is said, so people with limited English can get care. The company that makes the app handles patient information, so the doctor needs a signed agreement with it first.

A review of 20 studies on large language models in telehealth also lists translation as a way to make visits easier to reach.

Support between visits

A review of 31 studies, from 2015 to 2024, found that AI chat tools and digital helpers can improve mental health support. They can also help sort patients by how urgent their needs are. HHS gives another example in its business associate FAQ: a chatbot on a clinic's patient website that answers symptom questions, sends reminders and books appointments. That chatbot has to follow the same privacy rules as the clinic.

In one study of a talking AI system, users felt less alone and less stressed. The people who watched over the system also had less work to do.

Watching your health at home, and earlier screening

The same telemedicine review describes wearable devices with AI that watch your heart and breathing between visits. It also covers AI that screens eyes for diabetes damage and glaucoma. In skin image checks, AI reached accuracy close to that of experts. Most of these results come from a few medical areas, and many still need more testing in everyday care.

TaskWhat the AI doesWho decides
Visit notesWrites a draft from the conversationThe doctor checks and signs it
Possible causesLists options and warning signsThe doctor
Language helpTranslates for people with limited EnglishYour care team
Between visitsAnswers questions, sends reminders, helps sort urgencyThe doctor, for any medical choice
Home monitoringTracks wearable data and flags changesThe doctor acts on the alert

How does AI keep your information safe?

The privacy rules cover the companies behind these tools, not only the clinic. HHS says a company that handles your health information for a doctor must sign a business associate agreement. That is a contract. It says the company has to protect your data, and the company can be held responsible too. HHS's telehealth guidance says doctors must use companies that follow HIPAA and sign these contracts.

AI note-taking companies say they encrypt recordings and limit who can see them. Most delete the recording and the note after a set time. NPR's doctor says the company is responsible if the data leaks.

New rules are also on the way. ONC's HTI-1 rule asks certain health software that uses predictive AI to show doctors how the tool was built and tested, including whether it was checked for fairness. The FDA's draft guidance from January 2025 asks companies to check whether an AI medical device works equally well for different groups of people, and to keep watching it after launch. It is still a draft, so it is advice, not a law.

Teams that build these tools can check where data goes. Tigerless Labs publishes phi-boundary-gate, an open-source tool. It checks health information as it moves through an AI system, including messages, retrieved documents, tool output, requests sent to a model and logs. Each team sets its own rules for what is allowed, redacted or blocked. Every match still needs a person to review it, and the tool cannot make a system HIPAA compliant on its own.

What does good AI look like?

A 2026 review lists what the field needs. You can use it as a checklist for any company that sells these tools:

  • It has been tested on data it did not learn from.

  • Its results are checked over time, for each group of patients.

  • It is clear about how sure it is.

  • A human steps in when the AI is unsure.

The authors also see one promising role for AI: flagging things a doctor might have missed, while the doctor makes the final call.

What can you ask during a visit?

You can ask simple questions about the tools your care team uses. Stanford HAI has a longer list. Start with these five:

  1. Is the visit being recorded, written up or summarized by AI?
  2. Where are the recording and the note kept, and for how long?
  3. Does the company have a HIPAA agreement with the clinic?
  4. Can I say no, and does saying no change my care?
  5. Does a doctor check every AI note before it goes in my file?

NPR's doctor asks each patient first. If someone says no, she types notes on the computer while they talk. Most of her patients say yes.

When should you skip the AI and talk to a person?

Talk to a person when you need an exam, a treatment choice or emergency care. Emergencies come first. If you have chest pain, trouble breathing, signs of a stroke, or thoughts of hurting yourself, call emergency services or a crisis line right away.

For mental health, the American Psychological Association says chatbots and wellness apps can help a little, but they should not replace a licensed provider (APA).

If you want to record your own visit, know that a recording made in a phone app or sent to an AI service falls outside HIPAA. Read the app's privacy policy, and ask your doctor first, as Medscape explains.

What if the AI note gets something wrong?

Check the note after the visit. Stanford HAI suggests asking where you can see the transcript of your visit, and how to submit a correction if you spot a mistake. An AI tool can mishear a word, mix up a medicine name or a dose, or leave out something you said. Ask whether a person reviews the note before it goes into your record, and ask the office to fix any error you find.

Conclusion

AI helps telehealth most when it saves doctors time, gives them more to check, and helps people who struggle with language or distance. The best setups keep a licensed doctor in charge, a signed privacy agreement in place, and clear written rules about your data.

Before your next telehealth visit, ask how AI is used during the call. It takes a minute, and the answer tells you a lot.

For more on AI and health coverage, read Prior Authorization: How AI Helps You Prepare and Can AI Help Families Coordinate Their Health Insurance Coverage?.

FAQ

How is AI in telehealth regulated?

Mostly through privacy contracts and new rules. A company that handles your health information for a doctor must sign a HIPAA agreement. ONC's HTI-1 rule asks certain health software to show how its AI was built and tested. The FDA has draft guidance for AI medical devices. It is advice, not a law yet.

How does AI help a doctor during a visit?

It can write the visit note and suggest possible causes for the doctor to check. The doctor makes the decision. In a 2026 review, AI-suggested urgent care advice was rated best more often than the doctors' advice, 77.1% to 67.1%, but the studies are small.

Can I say no to an AI note-taker?

Yes. NPR's family doctor, Mara Gordon, says that when a patient says no to an AI note-taker, she types notes on the computer while the patient talks. Tell the office before the visit so they can plan for it.

Is an AI chatbot okay for mental health support?

The American Psychological Association says chatbots and wellness apps can help a little, but they should not replace a licensed provider. In an emergency, call emergency services or a crisis line.

Can I record my own visit?

No law says you cannot. Eleven states require everyone being recorded to agree: California, Florida, Illinois, Maryland, Massachusetts, Michigan, Montana, New Hampshire, Oregon, Pennsylvania and Washington. Ask your doctor first. A recording made in a phone app or sent to an AI service falls outside HIPAA.

Sources

  1. AI-Assisted Clinical Decision Support Versus Standard Teleconsultation in Adult Medicine: A Systematic Review
  2. Integrating Artificial Intelligence Into Telemedicine: Evidence, Challenges, and Future Directions
  3. Large language models and generative AI in telehealth: a responsible use lens
  4. Medscape: When Patients Want to Record Their Office Visit
  5. NPR: I'm a human doctor with human intelligence. But I sure do like my AI notetaker
  6. HHS: Guidance on audio-only telehealth under HIPAA
  7. HHS: Business Associates FAQs
  8. Telehealth.HHS.gov: HIPAA Rules for telehealth technology
  9. ONC: HTI-1 Final Rule
  10. FDA: Artificial Intelligence-Enabled Device Software Functions (draft guidance, January 2025)
  11. Stanford HAI: How Is AI Changing Your Doctor Visit?
  12. APA: Health advisory on generative AI chatbots and wellness apps

Share this article

Keep reading

11 min read

Should Agent Memory Have an Expiration Date?

A fixed expiry can drop facts that are still true. Here is how TTL, decay and event-based expiry differ, and when a fact should stay valid until replaced.

Read story