About 34% of US adults now use AI chatbots for health information, and Medicare’s ACCESS model, which started in July 2026, pays technology-enabled chronic care $90 to $420 per beneficiary per year. The FDA lists 1,614 AI-enabled medical devices, yet consumer chatbots sit outside HIPAA, and ChatGPT Health undertriaged 51.6% of true emergencies in a Nature Medicine test.
AI Health Apps in America 2026
AI Health Apps in US 2026 describe a market that grew faster than its rules. A Pew Research Center survey fielded in June 2026 found that 34% of US adults use AI chatbots for at least one health reason. Digital health startups raised $7.4 billion in the first half of the year. Medicare started paying for AI-supported chronic care in July, and the FDA’s list of AI-enabled devices passed 1,600. The consumer apps that people use most are mostly not on that list, because the FDA treats many of them as general wellness or decision support tools.
That gap runs through every section of this report. Users trade personal health data for speed, and 29% say they feel comfortable doing it. Independent tests show wide variation in how well chatbots spot emergencies. Federal guidance loosened in January 2026, and states passed their own rules. Each table below names its data period, because surveys, funding reports, and regulatory lists run on different calendars.
Interesting Facts About AI Health Apps in the US 2026
| Fact | Data (Period) |
|---|---|
| US adults using AI chatbots for health | 34% (Pew, June 2026) |
| US adults using AI for health in the past 30 days | 25%, about 66 million (Gallup and West Health, April 2026) |
| Digital health venture funding | $7.4 billion across 244 deals (H1 2026) |
| Medicare ACCESS payment range | $90 to $420 per beneficiary per year |
| ACCESS participants approved | More than 150 |
| Apps in the Medicare App Library | 5 live (June 2026) |
| FDA-listed AI-enabled devices | 1,614 (June 29, 2026) |
| ChatGPT Health undertriage of true emergencies | 51.6% (Nature Medicine, February 2026) |
| Users who skipped a provider visit because of AI advice | 14% (Gallup and West Health, April 2026) |
| State laws on AI in health or mental health | At least 9 enacted or scheduled |
These figures come from different surveys and agencies, so they do not add up to one picture. The Pew number covers the past year, and the Gallup number covers the past 30 days. The Medicare and FDA figures count programs and devices, not users. Each row stands on its own source, and the period in the right column matters.
The facts show an adoption curve ahead of oversight. One in three adults use chatbots for health, while only 5 apps had cleared the Medicare App Library by June. Funding hit $7.4 billion while the FDA’s January guidance eased the rules for decision support software. Regulators, insurers, and developers are all working from partial data, and that uncertainty is itself a finding.
AI Health App Usage in the US 2026
US adults using AI for health, by survey
Adults 18-29 (Pew, Jun 2026) ████████████████████████████████████████ 44%
All adults (Pew, Jun 2026) ███████████████████████████████ 34%
Physical health (KFF, Feb-Mar 2026) ██████████████████████████ 29%
Past 30 days (Gallup, Apr 2026) ███████████████████████ 25%
At least sometimes (Pew, Apr 2026) ████████████████████ 22%
Adults 65+ (Pew, Jun 2026) ███████████████ 17%
Mental health (KFF, Feb-Mar 2026) ███████████████ 16%
| Metric | Value (Source, Period) |
|---|---|
| Adults using chatbots for at least one health reason | 34% (Pew, June 2026, 3,488 adults) |
| Adults 18 to 29 using chatbots for health | 44% (Pew, June 2026) |
| Adults 65 and older using chatbots for health | 17% (Pew, June 2026) |
| Use AI for health in the past year, physical health | 29% (KFF, February to March 2026) |
| Use AI for health in the past year, mental health | 16% (KFF, February to March 2026) |
| Used AI for health in the past 30 days | 25% (Gallup and West Health, April 2026) |
| Use chatbots for health at least sometimes | 22%; 7% often (Pew, April 2026) |
| Chatbot users rating the answers extremely or very helpful | 47% (Pew, June 2026) |
| Users rating chatbot health answers highly accurate | 18% (Pew, April 2026) |
Data source: Pew Research Center (surveys of April 2026 and June 22 to 28, 2026); KFF health information and trust tracking poll (March 2026); West Health-Gallup Center on Healthcare in America (April 2026).
Adoption splits by age and survey wording. Pew’s June survey puts use at 34% for any health reason, and 44% among adults 18 to 29, nearly three times the 17% rate for adults 65 and older. KFF found the same age gap on mental health, where adults under 30 are three times as likely as adults 50 and older to use AI. Gallup’s 25% counts a shorter window, the past 30 days. The numbers differ because the questions differ, not because the surveys disagree.
Use and trust do not match. Only 18% of users in Pew’s April survey rated chatbot health answers highly accurate, yet 47% in June called the tools extremely or very helpful. Speed drives the choice: 28% of users cite getting information quickly, 22% cite low cost, and 20% use chatbots to interpret lab results. Gallup found that 46% of users felt more confident talking to a provider afterward, so most people use AI to prepare for care and not to replace it.
AI Health App Funding & Market Statistics 2026
US digital health venture funding, billions of dollars
H1 2025 ███████████████████████████████████ $6.4B
Q1 2026 ███████████████████████ $4.2B
Q2 2026 █████████████████ $3.2B
H1 2026 ████████████████████████████████████████ $7.4B
| Metric | Value (Period) |
|---|---|
| Digital health venture funding | $7.4 billion (H1 2026) vs $6.4 billion (H1 2025) |
| Deals | 244 (H1 2026) vs 245 (H1 2025) |
| Quarterly funding | $4.2 billion (Q1), $3.2 billion (Q2) |
| Mega-deals of $100 million or more | 20 rounds by 19 companies |
| Share of capital in mega-deals | 45% from just over 8% of deals |
| Median deal size | $14 million (2025: $12 million) |
| Digital health acquisitions | 115 (H1 2026); 71 in Q2 alone |
| Health systems with an “Epic-first” AI buying strategy | 71% |
Data source: Rock Health H1 2026 digital health funding and market overview (July 13, 2026); Redesign Health study of health system AI purchasing, as reported by HIT Consultant.
Funding rose $1 billion over the first half of 2025 on almost the same number of deals. That means bigger checks, not more companies. Twenty rounds of $100 million or more took 45% of all capital. Rock Health dropped its “AI-enabled” label in 2026 because AI is now too common to separate winners from the field. Investors now ask what a startup offers that a general AI model cannot.
Consolidation is visible on the buyer side. 115 acquisitions closed in the first half, and 71 of them landed in the second quarter, the busiest quarter since 2021. Health systems also lean on their electronic health record vendor: 71% follow an “Epic-first” purchasing rule, which makes it harder for stand-alone apps to win hospital contracts. For consumer apps, the practical effect is fewer independent products and more AI built into platforms users already open.
Medicare Payments for AI Health Apps Statistics 2026
ACCESS and Medicare App Library counts
Interest forms ████████████████████████████████████████ 350+
Approved participants █████████████████ 150+
App Library, expected soon █ 8
App Library, live (Jun 2026) █ 5
| Metric | Value (Period) |
|---|---|
| ACCESS model start | July 2026, runs for 10 years |
| Payment per beneficiary | $90 to $420 per year, tied to outcomes |
| Organizations that filed interest | More than 350 |
| Participants provisionally approved | More than 150 |
| Original conditions | Hypertension, diabetes, high cholesterol, chronic musculoskeletal pain, depression, anxiety |
| September 2026 expansion | 4 more conditions |
| Medicare App Library launch | April 2026 |
| Apps live in the Library | 5 (June 2026), 8 more expected |
| Payment structure | Monthly outcome-aligned payments; full amount only when patients meet goals |
Data source: Centers for Medicare & Medicaid Services Innovation Center, ACCESS Model; KFF analysis of CMS digital health efforts (June 2026); HFMA payment range as reported by MarketScale (September 2026); STAT News (April 2026).
ACCESS changes how Medicare pays for software. Instead of billing each telehealth or monitoring service, approved participants receive set annual rates between $90 and $420 per beneficiary, and they earn the full amount only when patients reach goals such as lower blood pressure or reduced pain. CMS officials said applications exceeded expectations, with more than 350 interest filings and more than 150 approvals. The low rates are deliberate: one participant told TechCrunch that rates must stay low to reward AI-driven care.
The Medicare App Library is a separate and smaller effort. It lists third-party apps that passed independent privacy and security review, and it held 5 apps in June, with 8 more expected. People in traditional Medicare can reach those apps through CMS. No public figure shows total Medicare spending on AI health apps yet, because ACCESS only began in July. Watch the first outcome reports for evidence on whether the model saves money.
Patient Data & Privacy in AI Health Apps Statistics 2026
Patient data attitudes, % of chatbot health users
Comfortable sharing health data ████████████████████████████████████████ 29%
Not comfortable sharing ████████████████████████████████████ 26%
Use chatbots for lab results ████████████████████████████ 20%
| Metric | Value (Period) |
|---|---|
| Users extremely or very comfortable sharing health data with a chatbot | 29% (Pew, June 2026) |
| Users not too or not at all comfortable | 26% (Pew, June 2026) |
| Users who interpret lab results with chatbots | 20% (Pew, June 2026) |
| People asking ChatGPT for health advice each day | About 40 million (OpenAI figure, reported March 2026) |
| ChatGPT Health launch | January 2026, links medical records and wellness apps |
| HIPAA coverage for consumer chatbots | None for standard consumer plans |
| Texas AI disclosure law | Effective January 1, 2026 |
| Adults who used AI for physical health and saw no clinician afterward | 42% (KFF, 2026) |
Data source: Pew Research Center (August 25, 2026); KFF (March 2026); Medical Economics reporting on OpenAI usage (March 2026); HIPAA Journal compliance guidance (2026).
Users are split on privacy. In Pew’s survey, 29% of chatbot health users feel extremely or very comfortable sharing personal health information, and 26% feel not too or not at all comfortable. People share anyway: about 40 million daily ChatGPT users ask health questions, by OpenAI’s account. ChatGPT Health lets users link medical records and wellness apps. HIPAA protects data held by providers and insurers, and consumer chatbot plans fall outside it, so a breach or misuse carries different legal consequences.
The gap affects what users can do after something goes wrong. A hospital patient has HIPAA rights to access, correction, and breach notice. A chatbot user relies on the company’s privacy policy and state law. Texas now requires disclosure when AI is used in clinical interactions, and 42% of adults who used AI for physical health advice did not follow up with a clinician. That combination, sensitive data plus no human check, drives most of the policy debate in 2026.
AI Health App Safety Statistics 2026
Undertriage of true emergencies, % of cases
ChatGPT Health, true emergencies ████████████████████████████████████████ 51.6%
OpenEvidence, true emergencies ██████████ 12.5%
| Safety Metric | Value (Source, Period) |
|---|---|
| ChatGPT Health undertriage of true emergencies | 51.6% (Nature Medicine, February 2026) |
| Test design | 60 scenarios, 21 specialties, 960 responses |
| OpenEvidence undertriage of true emergencies | 12.5% (medRxiv preprint, April 2026) |
| Accuracy gain in patient-style messages vs exam-style prompts | +6.4 points (Macquarie University, March 2026) |
| Users who received advice they believed unsafe | 11% (Gallup and West Health, April 2026) |
| Users who skipped a visit they would otherwise have made | 14%, about 14 million adults |
| Users who still saw a provider | 84% |
| Diagnostic errors that kill or disable Americans each year (clinicians) | 795,000 (AHRQ-funded study) |
Data source: Nature Medicine, Ramaswamy et al., Mount Sinai (February 23, 2026); medRxiv (April 2026); Macquarie University Centre for Health Informatics (March 2026); West Health-Gallup Center on Healthcare in America (April 2026); BMJ Quality & Safety.
The Mount Sinai team tested ChatGPT Health on 60 physician-reviewed scenarios across 21 specialties. It undertriaged 51.6% of true emergencies. It handled textbook cases such as stroke well, and it failed at the clinical extremes, including inconsistent suicide-crisis alerts. OpenAI said the study did not reflect typical use. A Macquarie University group found that patient-style messages raised accuracy by 6.4 points over the exam-style prompts. A physician-facing tool, OpenEvidence, undertriaged 12.5% under the same benchmark, in a preprint that has not yet been peer reviewed.
Real-world harm data come from surveys. Gallup found that 11% of recent users received advice they believed was unsafe, and 14%, about 14 million adults, skipped a provider visit because of AI advice. Still, 84% saw a provider. Human care has errors too: an AHRQ-funded study estimates 795,000 Americans die or become disabled from diagnostic errors each year, and readers can see that baseline in medical error statistics in the US. The comparison shows what AI must beat, and it also shows why users turn to it.
FDA AI-Enabled Device Statistics 2026
FDA AI-enabled device list, cumulative authorizations
End of 2025 ████████████████████████████████████ 1,451
March 2026 ██████████████████████████████████████ 1,524
June 29, 2026 ████████████████████████████████████████ 1,614
| Metric | Value (Period) |
|---|---|
| AI-enabled devices on the FDA list | 1,614 (June 29, 2026) |
| Cumulative total, end of 2025 | 1,451 |
| Cumulative total, March 2026 | 1,524 |
| Radiology share of the list | 76.2% |
| Devices authorized in 2025 | A record 331 (March 2026 update) |
| Devices already listed for first half of 2026 | 181 |
| FDA generative AI discussion paper | August 18, 2026 |
| First FDA list entry | 1995 |
| Pace of new authorizations | About 30 per month |
Data source: U.S. Food and Drug Administration, AI-Enabled Medical Device List (June 29, 2026 update); MedTech Dive; The Imaging Wire; Complizen analysis of the FDA list.
The FDA list passed 1,600 devices in 2026, and radiology holds 76.2% of it. Most listed products read images for clinicians, and few are patient-facing apps. The FDA says the list is not comprehensive: it flags devices by AI-related terms in authorization summaries, and it sometimes adds older devices later, which explains why trackers report different totals for the same year. Authorizations ran at about 30 a month, with 181 listed by June.
The most relevant item for app users is the August 18, 2026 FDA discussion paper on generative AI-enabled devices, which asks how to assess risk for tools built on large language models. The list does not yet identify LLM-based devices. For the wider clinical picture, see AI in healthcare statistics in the US, which tracks adoption in hospitals and clinics.
AI Health App Regulation & State Laws Statistics 2026
State laws on AI in health care, by effective year
2025 █████████████ 2 laws
2026 ████████████████████████████████████████ 6 laws
2027 ███████ 1 law
| Rule | Level | Status and Date |
|---|---|---|
| FDA General Wellness guidance (revised) | Federal | January 6, 2026 |
| FDA Clinical Decision Support guidance (revised) | Federal | January 6, 2026; reissued January 29, 2026 |
| Executive order on preempting state AI laws (EO 14365) | Federal | December 11, 2025 |
| Utah HB 452 (mental health chatbots) | State | May 7, 2025 |
| Nevada AB 406 (AI mental and behavioral health) | State | July 1, 2025 |
| California SB 243 (companion chatbots) | State | January 1, 2026 |
| Texas TRAIGA (AI disclosure in clinical care) | State | January 1, 2026 |
| Vermont Act 156, Rhode Island S 2197A and H 7538A | State | June 2026 |
| Tennessee SB 1580 (AI cannot pose as a mental health professional) | State | July 1, 2026 |
| Oregon SB 1546 (companion chatbots) | State | January 1, 2027 |
Data source: U.S. Food and Drug Administration guidance documents (January 2026); White House executive order of December 11, 2025; Live Compliance state AI health law tracker (2026); Imaging Wire reporting on Texas TRAIGA.
The FDA’s January 6 guidance moved in the permissive direction. It widened the general wellness category to cover non-invasive sensing of blood pressure, oxygen saturation, and glucose-related signals when marketed for wellness only. It also allows enforcement discretion for decision support software that offers a single clinically appropriate recommendation. Commentators note that the guidance stays largely silent on consumer symptom checkers and health chatbots, so the main consumer AI products still sit in a gray zone.
States moved the other way. At least 9 state laws now govern AI in health care or mental health, 6 of them effective in 2026. Most target mental health chatbots, companion bots, and AI that poses as a clinician. A December 2025 executive order directs the federal government to challenge state AI laws it calls onerous, so the patchwork may shrink. Courts also fill the gap: trackers followed 152 AI lawsuits as of October 6, 2026, including 6 product liability and wrongful death cases, and readers can review them in AI lawsuit statistics in the US.
Frequently Asked Questions About AI Health Apps in US 2026
How many Americans use AI for health information in 2026?
About 34% of US adults use AI chatbots for at least one health reason, per Pew’s June 2026 survey. Gallup found 25%, about 66 million adults, used AI for health in the past 30 days.
Does Medicare pay for AI health apps?
Partly. The CMS ACCESS model started in July 2026 and pays approved participants $90 to $420 per beneficiary per year for technology-supported chronic care, tied to outcomes.
What is the Medicare ACCESS model?
It is a 10-year CMS Innovation Center payment model for tech-enabled chronic care in Original Medicare. More than 150 participants were approved, and CMS added 4 more conditions in September 2026.
What conditions does ACCESS cover?
It started with hypertension, diabetes, high cholesterol, chronic musculoskeletal pain, depression, and anxiety. CMS expanded it by 4 conditions in September 2026.
What is the Medicare App Library?
It is a CMS directory of third-party health apps that passed independent privacy and security review. It launched in April 2026 and listed 5 apps in June.
Are AI health chatbots covered by HIPAA?
Standard consumer chatbot plans are not. HIPAA applies to providers, insurers, and their business associates, so data shared with a consumer chatbot lacks those legal protections.
How accurate are AI health chatbots?
Results vary. In a Nature Medicine test, ChatGPT Health undertriaged 51.6% of true emergencies, and a physician-facing tool undertriaged 12.5% in a preprint. Only 18% of users in Pew’s April survey rated answers highly accurate.
How many FDA-authorized AI medical devices are there?
The FDA’s list held 1,614 devices as of June 29, 2026, and radiology made up 76.2%. The FDA says the list is not comprehensive.
Has the FDA regulated AI chatbots?
Not directly. Its January 6, 2026 guidance relaxed oversight of decision support software and wellness products and says little about consumer health chatbots. The agency published a generative AI discussion paper on August 18, 2026.
Which states regulate AI in health care?
Utah, Nevada, California, Texas, Vermont, Rhode Island, Tennessee, and Oregon have passed laws, at least 9 in total. Most target mental health chatbots and AI that poses as a clinician.
Do people skip doctor visits because of AI advice?
Some do. Gallup found 14% of recent users skipped a provider visit they would otherwise have made, while 84% still saw a provider.
How much money is going into digital health in 2026?
Venture funding reached $7.4 billion across 244 deals in the first half of 2026, up from $6.4 billion a year earlier, per Rock Health.
Disclaimer: The data research report we present here is based on information found from various sources. We are not liable for any financial loss, errors, or damages of any kind that may result from the use of the information herein. We acknowledge that though we try to report accurately, we cannot verify the absolute facts of everything that has been represented.
