Vendor compliance profile
Is Claude HIPAA compliant?
Claude is not universally HIPAA-ready. Anthropic says it may provide a BAA for qualifying customers using specified HIPAA-eligible services, such as eligible first-party API deployments, but Claude.ai Free, Pro, Max, general Claude for Work, Workbench, Console, and other listed features are not covered by that BAA.
Direct compliance answer
Claude HIPAA, BAA, PHI, and SOC 2 snapshot
Last checked: 2026-10-01 | Confidence: High
| Direct answer | Claude is not universally HIPAA-ready. Anthropic says it may provide a BAA for qualifying customers using specified HIPAA-eligible services, such as eligible first-party API deployments, but Claude.ai Free, Pro, Max, general Claude for Work, Workbench, Console, and other listed features are not covered by that BAA. |
|---|---|
| BAA availability | Anthropic says it may provide a BAA after reviewing HIPAA-related requirements and the customer's use case. The public article identifies eligible first-party API use as an example and lists several products and features that are not covered. |
| Can it handle PHI? | Prompts, uploaded files, tool calls, outputs, caches, logs, evaluations, support cases, web search, batch jobs, connectors, and downstream applications can expose PHI outside eligible scope. |
| SOC 2 caveat | Anthropic maintains a public Trust Center with SOC 2 and other assurance signals. Review current artifact scope and confirm that it covers the exact Anthropic services and period under review. |
| What to verify | Whether Anthropic has approved the exact use case and executed a BAA for the correct API organization and HIPAA-eligible services. Which models, endpoints, regions, retention settings, logs, support paths, files, batch, prompt caching, web search, tools, and integrations are covered or excluded. |
Scope of this profile
Use this profile for Anthropic's product-specific BAA eligibility, first-party API scope, excluded Claude products and features, retention, tools, files, logging, and downstream AI data paths.
HIPAA status signal
Eligible services only
BAA public signal
Available after Anthropic review
SOC 2 evidence signal
Public trust evidence
PHI warning: Prompts, uploaded files, tool calls, outputs, caches, logs, evaluations, support cases, web search, batch jobs, connectors, and downstream applications can expose PHI outside eligible scope.
Search query answers
Is Claude HIPAA compliant?
Claude should not be treated as universally HIPAA compliant. Anthropic may offer a BAA for qualifying HIPAA-eligible commercial services, but product, endpoint, retention, feature, integration, and configuration scope must be confirmed before any prompt, file, output, or log contains PHI.
Does Anthropic offer a BAA for Claude?
Anthropic says it may provide a BAA after reviewing the customer's use case for certain HIPAA-eligible services, such as eligible first-party API use. Qualification can include zero data retention and other service-specific requirements.
Is Claude for Work covered by Anthropic's BAA?
Anthropic's current public BAA article says general Claude for Work Team and Enterprise plans are not covered, along with Claude.ai Free, Pro, Max, Workbench, Console, beta or chat products, and other listed features or integrations.
Can Claude process PHI through the API?
Potentially, but only after Anthropic approves the use case, the customer qualifies for a HIPAA-eligible service, a BAA is executed, and every model, endpoint, retention setting, file, tool, cache, batch, web-search, logging, and downstream data path remains within written scope.
HIPAA, BAA, and SOC 2 summary
| HIPAA | Anthropic publishes a limited HIPAA path for qualifying commercial customers using specified eligible services. This is not a blanket statement covering Claude.ai, Claude for Work, every API feature, or every integration. |
|---|---|
| BAA | Anthropic says it may provide a BAA after reviewing HIPAA-related requirements and the customer's use case. The public article identifies eligible first-party API use as an example and lists several products and features that are not covered. |
| SOC 2 | Anthropic maintains a public Trust Center with SOC 2 and other assurance signals. Review current artifact scope and confirm that it covers the exact Anthropic services and period under review. |
| PHI risk | Prompts, uploaded files, tool calls, outputs, caches, logs, evaluations, support cases, web search, batch jobs, connectors, and downstream applications can expose PHI outside eligible scope. |
| Category | HIPAA-Compliant AI Chatbots and Assistants |
| Last checked | 2026-10-01 |
| Confidence | High |
Public evidence and open questions
What public sources say
- Anthropic may provide a BAA for qualifying customers and specified HIPAA-eligible services after use-case review.
- Anthropic's public guidance says the BAA does not cover Claude.ai Free, Pro, or Max; general Claude for Work Team or Enterprise; Workbench; Console; and other listed beta, chat, feature, or integration paths.
- Anthropic notes that BAA-scoped customers may face configuration requirements and exclusions involving web search, batch processing, explicit prompt caching, or Files API uploads.
What remains unconfirmed
- Whether the buyer qualifies for Anthropic's BAA and which first-party API organization, models, endpoints, regions, support paths, and retention controls are included in the executed terms.
- Whether files, batch processing, prompt caching, web search, tools, connectors, evaluations, safety review, logs, support, and third-party integrations are covered or excluded.
- Whether the buyer's application, cloud provider, storage, observability, human review, and downstream systems have separate appropriate coverage.
What it may be used for
- Approved first-party API workflows after Anthropic confirms eligibility, executes a BAA, and documents product, feature, retention, and configuration scope.
- Healthcare AI applications that minimize PHI, govern prompts and outputs, control logging, and keep every cloud, storage, observability, and human-review path inside reviewed agreements.
- Vendor comparison against ChatGPT and AWS Bedrock where product-specific BAA and feature exclusions are evaluated rather than brand-level claims.
What not to use it for
- Entering PHI into Claude.ai Free, Pro, or Max, general Claude for Work, Workbench, Console, or another product excluded by Anthropic's public BAA guidance.
- Using web search, files, batch, prompt caching, tools, connectors, or third-party integrations with PHI unless the executed BAA explicitly covers the feature and workflow.
- Assuming Anthropic's SOC 2 evidence or an Enterprise subscription alone authorizes PHI use.
What to verify with the vendor
- Whether Anthropic has approved the exact use case and executed a BAA for the correct API organization and HIPAA-eligible services.
- Which models, endpoints, regions, retention settings, logs, support paths, files, batch, prompt caching, web search, tools, and integrations are covered or excluded.
- How the customer governs prompts, outputs, application logs, cloud hosting, observability, evaluations, human review, deletion, and downstream systems.
- Whether current Anthropic Trust Center evidence covers the exact service and operating period used in procurement review.
Safer alternatives and related profiles
Safer alternatives to consider
- ChatGPT or OpenAI API products only where the buyer qualifies for the relevant BAA and verifies eligible product, retention, connector, and feature scope.
- AWS Bedrock under the AWS BAA and current HIPAA Eligible Services Reference, with model, region, logging, agent, knowledge-base, and connected-service review.
- A healthcare-specific AI product with explicit BAA coverage when the complete clinical workflow, hosting, integrations, and human review need one accountable vendor path.
ChatGPT
HIPAA: Conditional | SOC 2: Public evidence
AWS Bedrock HIPAA model and BAA checklist
HIPAA: Conditional | SOC 2: AWS public evidence
HubSpot
HIPAA: Conditional | SOC 2: Public evidence
Klaviyo
HIPAA: Not supported for health data | SOC 2: Public trust signal
Wix
HIPAA: Conditional | SOC 2: Verify with vendor
FAQ
Is Claude HIPAA compliant?
Claude should not be treated as universally HIPAA compliant. Anthropic may offer a BAA for qualifying HIPAA-eligible commercial services, but product, endpoint, retention, feature, integration, and configuration scope must be confirmed before any prompt, file, output, or log contains PHI.
Does Anthropic offer a BAA for Claude?
Anthropic says it may provide a BAA after reviewing the customer's use case for certain HIPAA-eligible services, such as eligible first-party API use. Qualification can include zero data retention and other service-specific requirements.
Is Claude for Work covered by Anthropic's BAA?
Anthropic's current public BAA article says general Claude for Work Team and Enterprise plans are not covered, along with Claude.ai Free, Pro, Max, Workbench, Console, beta or chat products, and other listed features or integrations.
Can Claude process PHI through the API?
Potentially, but only after Anthropic approves the use case, the customer qualifies for a HIPAA-eligible service, a BAA is executed, and every model, endpoint, retention setting, file, tool, cache, batch, web-search, logging, and downstream data path remains within written scope.
Will Claude sign a BAA?
Anthropic says it may provide a BAA after reviewing HIPAA-related requirements and the customer's use case. The public article identifies eligible first-party API use as an example and lists several products and features that are not covered.
Can Claude be used with PHI?
Do not use this vendor with PHI until your organization verifies BAA scope, covered services, configuration, access controls, data retention, and connected integrations.
Does SOC 2 mean Claude is HIPAA compliant?
No. SOC 2 evidence can support security diligence, but it does not prove HIPAA compliance, confirm BAA coverage, or approve PHI use. Review HIPAA terms, BAA scope, covered services, configuration, and intended workflow separately.
What should buyers verify before using Claude with PHI?
Whether Anthropic has approved the exact use case and executed a BAA for the correct API organization and HIPAA-eligible services. Which models, endpoints, regions, retention settings, logs, support paths, files, batch, prompt caching, web search, tools, and integrations are covered or excluded. How the customer governs prompts, outputs, application logs, cloud hosting, observability, evaluations, human review, deletion, and downstream systems. Whether current Anthropic Trust Center evidence covers the exact service and operating period used in procurement review.
Last checked and source notes
- Last checked
- 2026-10-01
- Confidence
- High
- Dataset rows
- 274 vendors
- Reviewed Anthropic's BAA guidance for commercial customers and Anthropic Trust Center on 2026-10-01.
- Anthropic's public BAA guidance was updated in the week of this review and should be re-checked during contracting because eligible services and exclusions can change.
- ComplySaaS did not verify a private Anthropic BAA, approved customer use case, API organization, or deployment configuration.
- Anthropic BAA guidance for commercial customers
- Anthropic Trust Center