AI Provider Trust Registry evidence verified as of 2026-10-03

Registry / compare

Claude via AWS Bedrock vs Claude via Google Vertex AI

The same dimension can grade differently depending on who serves the model. Every cell links to its source; grades are evidence grades, not endorsements.

Dimension Claude via AWS Bedrock Claude via Google Vertex AI
SOC 2 Type II ◐Yes, sales-gated Audit confirmed: trust.anthropic.com is Anthropic's own Trust Center showing SOC 2 Type 2 in scope with the "Type 2 SOC 2 and CSA STAR L2 Report.pdf" resource, but sensitive... ◐Yes, sales-gated Anthropic’s Trust Center lists SOC 2 Type II coverage for Claude and the report is only available after a gated request, confirming sales‑gated availability.
ISO 27001 ●Yes, public Anthropic’s Trust Center publicly lists ISO 27001 certification as covering the model itself, which includes Claude when offered via AWS Bedrock. ●Yes, public Anthropic’s public Trust Center shows ISO 27001 certification with a note that it applies to the Claude model and partner‑supplied containers, thus covering Claude when offered...
ISO 42001 ●Yes, public Anthropic, the developer of Claude, publicly confirms ISO/IEC 42001 certification for its AI management system, which by policy covers Claude on all platforms including AWS Bedrock. ●Yes, public Anthropic's public ISO/IEC 42001 certification for its AI management system covers Claude, so the model on Google Vertex AI is certified.
Trust center ●Yes, public AWS Artifact is a publicly documented, self‑service compliance portal that all AWS accounts can access, confirming a maintained trust center. ●Yes, public Google Cloud's own trust center (cloud.google.com/trust-center) is a publicly accessible, maintained compliance portal covering Google Cloud products including Vertex AI, and...
HIPAA BAA ◐Yes, sales-gated AWS lists Amazon Bedrock as HIPAA‑eligible but requires customers to first sign an AWS Business Associate Agreement, which is only available through the gated AWS Artifact portal. ●Yes, public Google's own HIPAA page and BAA support page publicly offer a self-serve Business Associate Addendum (review-and-accept in the Cloud Console), covering Google Cloud...
GDPR DPA ●Yes, public AWS provides a publicly accessible GDPR Data Processing Addendum that includes SCC language and references a publicly listed sub‑processor page, covering all AWS services... ●Yes, public The DPA page confirms a public DPA with SCCs, and Google provides a publicly accessible subprocessor list, satisfying both requirements.
No-training default ●Yes, public The AWS Bedrock FAQ publicly commits that customer content is not used to improve base models, confirming a default no‑training‑on‑API‑data policy for Claude via Bedrock. ●Yes, public The provider's public data governance page explicitly states a commitment not to use customer data for training by default, and Claude is offered as a managed model on the...
Retention / ZDR ◔Partial AWS docs confirm retention settings are publicly documented and state that ZDR applies only to pre‑Fable 5 Claude models, so the claim is only partially satisfied. ●Yes, public Google publicly documents its zero-data-retention policy for all managed models on Gemini Enterprise Agent Platform, and Claude is listed as a managed model on the platform, so...
Residency ●Yes, public AWS Bedrock model card for Claude Sonnet 5 publicly lists an EU inference profile that pins data to EU regions. ●Yes, public Google Cloud’s blog confirms publicly that EU multi‑region endpoints let users pin data processing to the EU geography for Claude on Vertex AI.
GPAI Code ●Yes, public Anthropic is listed among the signatories on the EU Commission's official GPAI Code of Practice signatory page, confirming the model developer is a signatory. ●Yes, public The European Commission's official GPAI Code of Practice signatory list includes Anthropic, confirming the model developer is a signatory.
Art. 53 summary ◐Yes, sales-gated Anthropic’s Trust Center page for the Claude Sonnet 5 training‑data summary is gated behind a request‑access form, confirming the summary is not publicly available. ◐Yes, sales-gated Anthropic’s own Trust Center page for the Claude Sonnet 5 training‑data summary is gated behind a request‑access form, so the summary is not publicly available.