Claude can be used with patient information, but not on the plans most people sign up for. Anthropic offers a Business Associate Agreement (BAA) for its enterprise and developer products, and as of mid-2026 eligible organizations can accept it themselves without a sales call. Coverage then applies only to the features Anthropic lists. Here’s what qualifies as of November 2026 and how to set it up.
Which Claude plans can be HIPAA covered
| Claude product | BAA? | How |
|---|---|---|
| Claude Free, Pro, Max, Ultra | No | Individual plans can’t enable HIPAA. |
| Claude Team | No | Business plan, but can’t enable HIPAA. |
| Claude Enterprise | Yes | Primary Owner enables HIPAA in settings and accepts the BAA. |
| Claude API (Anthropic) | Yes | Accept the BAA in the Claude Console, or negotiate one through sales. |
| Claude in Microsoft Foundry | Yes | BAA with Anthropic plus HIPAA configuration on the Azure subscription. |
| Claude in Amazon Bedrock | Yes, through AWS | Covered by your AWS BAA. Bedrock is HIPAA eligible. |
| Claude Platform on AWS | No | AWS says it isn’t HIPAA eligible. Use Bedrock instead. |
| Claude on Google Cloud | Check first | Google’s BAA covers its generative AI platform; confirm partner-model coverage with Google. |
Anthropic’s support page is direct about this: “Team plans and individual plans (Free, Pro, Max, Ultra) can’t enable HIPAA.”
Turning it on: Claude Enterprise
On Enterprise, HIPAA isn’t on by default. Anthropic says standard Enterprise plans “do not include BAA coverage without action from a Primary Owner.” To enable it, the Primary Owner goes to Organization settings → Data and privacy → HIPAA Compliance, reviews the BAA and implementation guide, and selects Accept and enable HIPAA. That click is acceptance of the BAA.
Two things to know before you click:
- It’s one-way. Anthropic describes it as a decision that can’t be reversed in settings, and enabling it resets some organization settings.
- Old API BAAs don’t carry over. A BAA signed for the API before December 2, 2025 covers only the API, not your Enterprise plan.
Turning it on: the Claude API
For developers building on the API, an authorized representative accepts the BAA in the Claude Console under Settings → Data retention → HIPAA compliance. Once accepted, the setting is permanent, and the API rejects requests to features that aren’t eligible with an error rather than silently processing them. HIPAA is set per organization, so keep PHI work in its own organization, separate from everything else.
What’s covered and what isn’t
On Claude Enterprise, Anthropic’s privacy center lists these as covered once HIPAA is enabled: chat, Projects, Artifacts in chat, file creation and code execution (without network access), voice, web search and Research. Not covered: connectors and MCP integrations, Claude in Chrome, Skills, Cowork in the cloud, and beta products.
On the API, eligible features include the Messages API with prompt caching, structured outputs, memory, the web search tool, the bash and text-editor tools, token counting, inline PDFs and the 1M-token context. Not eligible: the Batch API, the Files API, Skills, code execution, web fetch, the browser-use tool, external MCP, Managed Agents and most beta features.
Anywhere: no PHI in workspace names, user profiles, billing details or support tickets, and no PHI typed into the Claude Console interface itself.
Claude Code is the murky one. Anthropic’s API documentation says Claude Code is “not covered under HIPAA readiness,” while its privacy center describes coverage on Enterprise with zero data retention or a HIPAA configuration that’s currently only available through an Anthropic account team. If developers will work near PHI, get the arrangement confirmed in writing.
Claude through AWS, Microsoft and Google
You may not need Anthropic’s BAA at all. If your business already has a BAA with a cloud provider, you can use Claude through that provider instead.
- Amazon Bedrock. Bedrock is on AWS’s HIPAA-eligible services list, so Claude on Bedrock falls under your AWS BAA. AWS says Bedrock data “is not shared with model providers, and is not used to improve the base models.” Anthropic’s docs describe the cloud provider as the data processor here.
- The trap: Claude Platform on AWS. This is a different product, operated by Anthropic and sold through AWS. AWS’s own documentation says Anthropic’s HIPAA-ready program isn’t available on it and that customers with HIPAA requirements “should evaluate Claude in Amazon Bedrock instead.”
- Microsoft Foundry. Claude in Foundry is eligible, but it needs a BAA with Anthropic and the HIPAA configuration on your Azure subscription, arranged through Anthropic’s account team.
- Google Cloud. Google’s HIPAA list includes its generative AI platform, but it doesn’t name partner models such as Claude. Confirm coverage with Google before you rely on it.
A note on newer models and retention
Anthropic now designates its newest models (Claude Fable 5 and 5.1, Claude Mythos 5 and 5.1) as “Covered Models” that require 30-day data retention, so zero data retention isn’t available for them unless Anthropic expressly authorizes it. On Bedrock, these models keep prompts and completions within AWS for up to 30 days, where AWS may review them, without sharing them with Anthropic. This doesn’t break HIPAA (retention under a BAA is allowed), but it matters if your policy requires zero retention.
What a BAA doesn’t do for you
Anthropic covers its side. You still own the rest of the HIPAA Security Rule: a risk analysis that includes Claude, access controls and single sign-on, minimum-necessary use, staff training, device security, audit logging, and a breach response plan. And if staff have been pasting patient details into personal Pro accounts or a Team workspace, treat that as a possible breach: under 45 CFR 164.402, it’s presumed to be one unless a documented risk assessment shows otherwise.
Choosing a setup
- A practice that wants a chat assistant for staff: Claude Enterprise with HIPAA enabled, with connectors left off.
- A business building Claude into its own systems (intake summaries, document review, patient messaging drafts): the API with a BAA, or Claude on Bedrock inside your AWS account.
- A business that wants patient data never to leave its own cloud: Bedrock in your own AWS account, connected to your EHR and document storage.
The last option is what we set up for clients: private AI in an account the business owns, wired into the tools it already uses. See AI consulting or start with the AI audit. For ChatGPT, Gemini and Copilot side by side, read our guide to HIPAA-compliant AI, or see is ChatGPT HIPAA compliant.
This is general information, not legal advice. Anthropic updates its covered-feature list often; check it on the day you configure.
Sources
- Anthropic privacy center, BAAs for commercial customers: https://privacy.claude.com/en/articles/8114513-business-associate-agreements-baa-for-commercial-customers
- Claude support, HIPAA-ready Enterprise plans: https://support.claude.com/en/articles/13296973-hipaa-ready-enterprise-plans
- Claude API documentation, API and data retention (HIPAA readiness): https://platform.claude.com/docs/en/manage-claude/api-and-data-retention
- Anthropic privacy center, data retention for Covered Models: https://privacy.claude.com/en/articles/15425996-data-retention-practices-for-covered-models
- Anthropic, Claude for Healthcare and Life Sciences: https://www.anthropic.com/news/healthcare-life-sciences
- AWS HIPAA eligible services reference: https://aws.amazon.com/compliance/hipaa-eligible-services-reference/
- Amazon Bedrock security and compliance: https://aws.amazon.com/bedrock/security-compliance/
- Amazon Bedrock data retention: https://docs.aws.amazon.com/bedrock/latest/userguide/data-retention.html
- AWS, Claude Platform on AWS feature support: https://docs.aws.amazon.com/claude-platform/latest/userguide/feature-support.html
- Google Cloud HIPAA compliance: https://docs.cloud.google.com/docs/security/compliance/hipaa
- 45 CFR 164.402 (breach definition): https://www.law.cornell.edu/cfr/text/45/164.402