On this page
Claude for Healthcare is Anthropic's set of tools, connectors and skills, announced on January 11, 2026, that lets healthcare providers, payers and health tech companies use Claude “for medical purposes through HIPAA-ready products” (Anthropic). It connects Claude to the CMS Coverage Database, ICD-10 and the NPI Registry, and adds skills for FHIR development and prior authorization review. Teams use it safely by keeping patient data in a HIPAA-ready plan with only covered features switched on, and by giving connected tools data that carries no patient details. For marketing questions, Curve MCP from Curve Compliance connects Claude to your own campaign and conversion results and answers with aggregate figures, so your team gets the numbers without the patient data.
Book a call. Want Claude to answer questions about your ad campaigns? Curve MCP connects it to your Curve Compliance results, and every answer is checked before it leaves Curve. The Curve team switches it on for your organization. Book a call with Curve.
What is Claude for Healthcare?
Anthropic launched Claude for Healthcare alongside new life sciences tools. In its words, it is “a complementary set of tools and resources that allow healthcare providers, payers, and health tech companies and startups to use Claude for medical purposes through HIPAA-ready products” (Anthropic). On February 7, 2026, Anthropic edited that introduction “to clarify that HIPAA-ready products are specific to healthcare providers and payers.” Organizations reach it through business plans: “Healthcare customers can access Claude via Claude for Enterprise and the Claude Platform” (Claude).
The package has three parts: connectors to healthcare databases, Agent Skills for common workflows, and HIPAA-ready access through Claude for Enterprise and the API. Its healthcare page describes Claude as “built on HIPAA-ready infrastructure with safety guardrails designed for healthcare” (Claude), with answers that can be traced back to their source.
What do the Claude for Healthcare connectors and skills do?
Connectors give Claude direct access to other platforms. Anthropic added these for payers and providers (Anthropic):
- CMS Coverage Database. Local and National Coverage Determinations, so Claude can “verify locally-accurate coverage requirements, support prior authorization checks, and help build stronger claims appeals.”
- ICD-10. Diagnosis and procedure code lookups for medical coding, billing accuracy and claims management, with data provided by CMS and the CDC.
- NPI Registry. Help with “provider verification, credentialing, networking directory management, and claims validation.”
- PubMed. Already available, and open to HIPAA-compliant organizations on Claude for Enterprise, with “more than 35 million pieces of biomedical literature.”
Two Agent Skills came with it: FHIR development, for connecting healthcare systems, and a sample prior authorization review skill that organizations can adapt to their own policies. Availability and BAA coverage are separate questions. Anthropic says “Our new connectors and Agent Skills are generally available to all Claude subscribers, including Claude Pro, Max, Teams, and Enterprise” (Anthropic), while its BAA article lists skills as “Off by default in organizations that have enabled HIPAA” (Anthropic's BAA article).
What can healthcare teams do with Claude?
Anthropic gives four main uses (Anthropic):
- Prior authorization reviews. Claude can “pull coverage requirements from CMS or custom policies, check clinical criteria against patient records in a HIPAA-ready manner, and then propose a determination with supporting materials for the payer’s review.”
- Claims appeals. Claude pulls together patient records, coverage policies, clinical guidelines and prior documentation, so providers build stronger appeals and payers process them faster.
- Care coordination and message triage. Claude helps care teams work through “a large volume of patient portal messages, referrals, and handoffs” and flags what needs attention first.
- Building healthcare products. Startups on the Claude Developer Platform can build tools such as ambient scribing for clinical documentation and chart review support.
Most of these touch patient records, which is why the setup in the next section matters more than the model. Claude's healthcare page frames the same work around coverage policies, coding systems and clinical guidelines, with sample skills that teams adapt to their own policies and data.
How do healthcare teams set Claude up safely?
Start with the plan and its settings. Anthropic's BAA covers its HIPAA-ready services, “such as use of our first-party API, Claude in Microsoft Foundry, or Enterprise plans,” but it isn't on by default: “Standard Claude Enterprise plans do not include BAA coverage without action from a Primary Owner,” who activates HIPAA compliance in the organization settings under Data and privacy and accepts the BAA (Anthropic's BAA article). For the API, “your organization's Primary Owner will need to sign a BAA and then reach out to your Anthropic contact or our Sales team to get this turned on.” The agreement “only covers the single organization that accepted it,” so each organization that handles patient data sets this up for itself.
Then decide which features your team may use with patient data. This is how Anthropic's BAA article sorts the main surfaces:
| Claude surface | Under Anthropic's BAA? | What Anthropic says |
|---|---|---|
| Chat, plus projects and artifacts in chat | Covered, after the Primary Owner step | Covered as Eligible Services; projects and artifacts count “in chat only” |
| Web Search and Research | Covered | Covered as Eligible Services |
| First-party API and Claude in Microsoft Foundry | Covered for listed features | Named among Anthropic's HIPAA-ready services, but “Not all API features are covered” |
| Skills | Not covered | Available, but “Off by default in organizations that have enabled HIPAA” |
| Connectors and MCP servers | Not covered for data sent out | “sending data to 3rd parties via this feature isn’t covered under Anthropic’s BAA” |
| Claude in Chrome | Not covered for data sent out | Off by default in HIPAA organizations |
| Claude Console and Cowork in the cloud | Excluded | Named in the BAA exclusions |
| Claude Design, Claude Slides and Claude Docs (beta) | Not covered | Beta features; “Not covered under BAA” |
| Claude Free, Pro, Max and Ultra | Not named as HIPAA-ready | The BAA article is “about our commercial products such as Claude for Work and the Anthropic API” |
The BAA version matters too. Anthropic marks the covered Enterprise features “Covered under versions of the BAA accepted after 12/2/25” and the covered Messages API features “Covered under versions of the BAA accepted after 4/1/26” (Anthropic's BAA article). Claude Code and Cowork running on a member's own computer can come under the BAA too, once the Primary Owner applies Anthropic's HIPAA configuration to both products. Tell your team to keep patient details out of every feature marked not covered. For the full verdict on each route to Claude, including Amazon Bedrock, read Is Claude HIPAA compliant? BAA options explained; the short vendor answer is at does Claude sign a BAA, and other assistants are compared in HIPAA compliant AI tools that sign a BAA.
What about connecting personal health records to Claude?
That is a separate feature for individuals. “In the US, Claude Pro and Max plan subscribers can choose to give Claude secure access to their lab results and health records,” through HealthEx and Function connectors and Apple Health and Android Health Connect integrations, all introduced in beta (Anthropic). Anthropic says users “must explicitly opt-in to enable access,” and adds: “We do not use users’ health data to train models.”
The feature lets a person read and understand their own records. It doesn't turn a staff member's Pro or Max account into a place for a clinic's patient data, because Anthropic's BAA article is written for commercial products. For a practice, the question stays the same: which commercial Claude service holds the data, and whether the feature in use is covered.
How should a marketing team connect Claude to ad data?
With aggregate results, and most marketing questions need nothing more. HIPAA treats a vendor that handles PHI for a clinic as a business associate. HHS lists “data analysis” among the activities a business associate performs, and names a “Cloud service provider engaged to create, receive, maintain, or transmit electronic PHI (ePHI)” as an example (HHS). A lead list with names and the treatment each person asked about, a call log or a form export can be PHI when a clinic holds it, so it belongs only on a covered Claude service, in a covered feature.
Connectors are the part marketing teams miss. Anthropic says connectors and MCP servers are “Available to use but sending data to 3rd parties via this feature isn’t covered under Anthropic’s BAA” (Anthropic's BAA article), so whatever a connector sends falls outside that agreement. Sort each source before you click allow: campaign KPIs, spend and weekly conversion counts are the safe tier, while contact records, form responses and call transcripts stay out. Connecting Claude to clinic ad data: what to allow sets out that policy in full, and approving an MCP connector gives the review steps. Write the policy down, and talk to your counsel about any connection that would carry patient details.
How Curve Compliance helps
Curve MCP connects Claude, ChatGPT, Cursor and any MCP client to your campaign and conversion results in Curve Compliance. Ask in plain English how your Meta or Google Ads campaigns did, and the answer covers visitors, sessions, goal completions, funnel steps and each campaign's spend, clicks and impressions next to the conversions Curve tracked. Curve MCP answers with aggregate results, every answer is checked before it leaves Curve, and every request is recorded with what was returned, so your compliance team can see exactly what was asked and answered. Ask about an individual patient and you get a single-use, short-lived link into Curve instead of data. Curve MCP brings your results into the AI client you already use, and Curve Analyst works inside Curve with full access to your data. See Curve MCP and Curve Analyst: when to use which.
The numbers come from Curve's HIPAA-compliant tracking. One script replaces browser ad pixels, conversions go server-side to Meta, Google Ads, TikTok, Microsoft Advertising and LinkedIn, and Curve strips PHI before data reaches an ad platform. Curve signs a BAA on every plan, is SOC 2 Type II, and its team does the setup for you, typically live in about a week. Book a call to switch on Curve MCP for your team.
Sources, checked November 20, 2026: Anthropic, Advancing Claude in healthcare and the life sciences; Claude, Healthcare solutions; Anthropic Privacy Center, BAAs for commercial customers; HHS, Business Associates FAQs.
Frequently Asked Questions
What is Claude for Healthcare?
Claude for Healthcare is Anthropic's set of connectors, skills and HIPAA-ready products that lets healthcare providers, payers and health tech companies use Claude for medical work. It launched on January 11, 2026, with connectors to the CMS Coverage Database, ICD-10 and the NPI Registry.
Who can use Claude for Healthcare?
Anthropic says its HIPAA-ready products are specific to healthcare providers and payers, reached through Claude for Enterprise and the Claude Platform (Anthropic; Claude). The new connectors and Agent Skills themselves are available to all Claude subscribers.
What does Claude for Healthcare connect to?
The CMS Coverage Database, ICD-10 and the NPI Registry, plus PubMed for HIPAA-compliant organizations on Claude for Enterprise (Anthropic). It also adds Agent Skills for FHIR development and prior authorization review.
What must be switched on before Claude handles patient data?
On Claude Enterprise, the Primary Owner activates HIPAA compliance under Data and privacy and accepts Anthropic's BAA, because “Standard Claude Enterprise plans do not include BAA coverage without action from a Primary Owner” (Anthropic's BAA article). For the verdict on each route to Claude, see Is Claude HIPAA compliant? BAA options explained.
Can Claude Pro or Max users connect their own health records?
Yes, in the US, through HealthEx, Function, Apple Health and Android Health Connect, all in beta (Anthropic). It is a personal feature for reading your own records, not a clinic workspace.
Does Anthropic train on health data?
For connected personal health records, Anthropic says: “We do not use users’ health data to train models” (Anthropic). For a business plan, the data terms are in your agreement with Anthropic.
Are Claude connectors covered by Anthropic's BAA?
No. Anthropic says sending data to third parties through connectors and MCP servers isn't covered under its BAA, even after an Owner turns them on (Anthropic's BAA article). Check what each connection sends before you allow it.
Can Claude read my ad results without patient data?
Yes, through Curve MCP. It connects Claude to your Curve Compliance campaign and conversion results, answers with aggregate results checked before they leave Curve, and records every request.
Talk to Curve Compliance
Book a call. Want to ask Claude how your campaigns did without handing it patient data? Curve MCP answers from your own Curve Compliance account, and Curve signs a BAA on every plan. Curve's team does the setup, typically live in about a week. Book a call with Curve.