Amazon Bedrock

The document exists. The ledger fetched it at its publisher and quotes it.

Record ID

AVD-2026-0026

Kind

API and platform

Jurisdiction

United States

Last verified

15 September 2026

Added

15 September 2026

• Amazon Bedrock uses a zero data retention security model, so by default it does not store model inputs or outputs.
• Where a model requires retention, prompts and completions are held within the AWS boundary for up to 30 days and are not shared with the model provider.
• Model providers have no access to the deployment accounts, so they see neither logs nor customer prompts and completions.
• Retention is a mode set at account or project level, and an organisation can require the zero retention mode through a service control policy.

Used for training by default, Retention period, Deletion on request, Storage region and residency, Subprocessor list published, Human review of content, Opt out available, DPA available, Persistent memory across sessions, Last change to the terms. Stated means the document you can open below says it; silent means the ledger read the document and it does not.

Used for training by defaultStated

Providers are structurally excluded. They have no access to the deployment accounts and therefore none to Bedrock logs or to customer prompts and completions.AWS documentation, Amazon Bedrock data protection, primary source, 15 September 2026.

Retention periodStated

Zero by default, with named models requiring up to 30 days for abuse detection, and per feature behaviour where the customer chooses to store responses.AWS documentation, Amazon Bedrock data retention, primary source, 15 September 2026.

Deletion on requestSilent

Expressed as an automatic cap rather than a deletion right. No customer initiated deletion of retained abuse data is described.AWS documentation, Amazon Bedrock data retention, primary source, 15 September 2026.

Storage region and residencyStated

Retained inputs and outputs are stored in the destination region where the request is processed, and a model deployment account exists per provider per region.AWS documentation, Amazon Bedrock abuse detection, primary source, 15 September 2026.

Subprocessor list publishedStated

The model providers are the third parties and they are walled off. Bedrock does not share customer content with model providers today.AWS documentation, Amazon Bedrock data retention, primary source, 15 September 2026.

Human review of contentStated

Only where a provider requires it as a condition of access, performed by AWS inside the AWS boundary, and the content does not leave AWS.AWS documentation, Amazon Bedrock data retention, primary source, 15 September 2026.

Opt out availableStated

Zero retention is set by API and can be enforced across an organisation by policy, and for models that require retention it is evaluated per account and per model.AWS documentation, Amazon Bedrock data retention, primary source, 15 September 2026.

DPA availableSilent

The Bedrock data protection pages point to the service terms and a regulatory centre rather than naming a data processing addendum.AWS documentation, Amazon Bedrock data protection, primary source, 15 September 2026.

Persistent memory across sessionsStated

Persistence is a request level choice through the responses API, and the documentation warns that setting store to false does not guarantee zero retention.AWS documentation, Amazon Bedrock data retention, primary source, 15 September 2026.

Last change to the termsStated

The Bedrock user guide pages print no date. The AWS service terms print Last Updated 15 September 2026.AWS service terms, primary source, 15 September 2026.

1. 30 days

   Retention cap for models that require review

   AWS documentation, Amazon Bedrock data retention, primary source, as of 15 September 2026.

2. 30 days

   Abuse detection retention for flagged traffic

   AWS documentation, Amazon Bedrock abuse detection, primary source, as of 15 September 2026.

Retention here is a property of the model you call, not of the platform you bought. A single account can be zero retention for one model and thirty days with human review for another, because each model declares the modes it allows, and the more permissive setting is what unlocks certain models at all. Any statement you make to a customer about retention has to name the models in the deployment, not the service.

1. AWS documentation, Amazon Bedrock data protectionPrimary · 15 September 2026
2. AWS documentation, Amazon Bedrock data retentionPrimary · 15 September 2026
3. AWS documentation, Amazon Bedrock abuse detectionPrimary · 15 September 2026
4. AWS service termsPrimary · 15 September 2026

• VerifiedAPI and platformAVD-2026-0004

  The OpenAI API and platform

  United States · verified 15 September 2026

  The platform guide states that as of 1 March 2023, data sent to the OpenAI API is not used to train or improve OpenAI models.

• VerifiedAPI and platformAVD-2026-0007

  The Anthropic API and Claude Developer Platform

  United States · verified 15 September 2026

  The API retention page states that retained data is never used for model training without your express permission.

• Open questionOpen questionAVD-2026-0030

  Does zero data retention survive an abuse investigation?

  Global · verified 15 September 2026

  OpenAI states a flagged image is retained for manual review even where zero data retention, modified abuse monitoring or eyes off is enabled.

• VerifiedBusiness tierAVD-2026-0025

  GitHub Copilot Business and Enterprise

  United States · verified 15 September 2026

  GitHub states it does not use Copilot Business or Copilot Enterprise customer data to train AI models.

• VerifiedBusiness tierAVD-2026-0024

  Cursor and its privacy mode

  United States · verified 15 September 2026

  The terms of service state that Cursor will not use content to train any AI models unless you have explicitly agreed.

• VerifiedBusiness tierAVD-2026-0023

  Perplexity enterprise and the Sonar API

  United States · verified 15 September 2026

  The developer documentation states Perplexity does not use customer data to train its models or for any purpose beyond processing the immediate request.