Anthropic security review: SOC 2, AI training & subprocessors

What Anthropic publishes about its security, privacy and use of your data, with a link to the page each fact came from. Checked October 10, 2026.

Transparency score
100/100
SOC 2
SOC 2
ISO 27001
Not mentioned
Trains AI on your data?
See quote
Subprocessors
List published

"Not found" means our reader didn't find it on the public pages it could read; the vendor may still publish it.

Does Anthropic train AI on your data?

Mentions using data for AI, with an opt-out

Anthropic's own words (anthropic.com/legal/privacy):

We may use your Inputs and Outputs to train and improve Anthropic AI models, unless you opt out through your account settings.

Vendors often have different rules for consumer or free plans and for business or enterprise plans. Check the terms for the plan you would use.

Certifications Anthropic claims

Certification or reportMentioned on
SOC 2anthropic.com/legal/data-processing-addendum
HIPAAanthropic.com/legal/data-processing-addendum

Privacy laws and frameworks mentioned: GDPR. These are laws the vendor says it follows, not certifications.

A certification on a website is a claim. For anything sensitive, ask Anthropic for the SOC 2 report or ISO certificate.

Anthropic subprocessors

Anthropic publishes its subprocessor list at trust.anthropic.com/subprocessors, but it couldn't be read automatically. Open it directly.

Security features mentioned

SSO / SAMLMentioned
SCIM provisioningNot found
Multi-factor authenticationMentioned
Encryption at restMentioned
Encryption in transitMentioned
Penetration testingMentioned
Bug bounty / disclosure programNot found
Audit logsNot found
Data residencyMentioned United States, UK, Ireland, Switzerland
security.txt contacthttps://hackerone.com/4f1f16ba-10d3-4d09-9ecc-c721aad90f24/embedded_submissions/new

Key documents

Trust / security centertrust.anthropic.com
Privacy policyanthropic.com/legal/privacy
Subprocessor listtrust.anthropic.com/subprocessors
Data processing agreement (DPA)anthropic.com/legal/data-processing-addendum
AI policyNot found
TermsNot found
Status pagestatus.anthropic.com

How this score is worked out

CheckPointsResult
Security or trust page15Yes
Certifications listed20Yes
Privacy policy10Yes
Subprocessor list15Yes
DPA available10Yes
AI training policy stated15Yes
security.txt contact5Yes
Status page10Yes

The score measures how much of a standard security review our reader could answer from Anthropic's public pages. It is not a rating of how secure Anthropic is, and a "No" can mean the document exists where our reader couldn't see it.

Check another vendor

Get the same profile for any vendor in seconds, or let your AI agent do it with Stormap's free MCP server.

Other AI vendors

Cohere · Copy.ai · Cursor · Descript · ElevenLabs · Glean · Harvey · Hugging Face · Jasper · Mistral AI · Pinecone · Replit

Source: Anthropic's own public website (5 pages read on October 10, 2026). Stormap is not affiliated with Anthropic. Facts are what the vendor publishes about itself; certifications are claims, not verified audit reports. Something wrong or out of date? Pages are re-checked every two weeks.