Does Amazon Bedrock store your API prompts?

Default retention

Depends on the model; up to 30 days for some

Per-Region data_retention_mode (none, default, aws_review, legacy provider_data_share, inherit). Under default, the model's policy applies; AWS may retain for abuse prevention for specific models: OpenAI models (except gpt-oss family) retain classifier-flagged traffic up to 30 days; Claude Fable 5/5.1 retain all traffic up to 30 days. No change for Claude models released before Fable 5.

“default | Default means the data retention policy of the model applies. There is no change to previous model retention behavior; if ZDR applied previously, then ZDR still applies.”

“For all OpenAI models other than GPT OSS Safeguard 120B, GPT OSS Safeguard 20B, gpt-oss-120b, and gpt-oss-20b, classifier-flagged traffic will be retained for up to 30 days for automated offline abuse detection. Eligible customers may request full ZDR through their AWS account team.”

“For Anthropic Claude Fable 5 and Claude Fable 5.1, all traffic will be retained for up to 30 days for automated offline abuse detection. Classifier-flagged traffic will be subject to potential human review performed by AWS.”

“There is no data retention change to Claude models released before Claude Fable 5.”

Duration

Model-dependent; up to 30 days for listed models

Purpose

automated offline abuse detection; human review by AWS where the provider requires it

Availability

Self-serve: you can switch it on yourself

data_retention_mode none, per Region; some models require retention

Who can get it

Self-serve data_retention_mode: none at account (per Region) or project level, enforceable with IAM/SCPs. Models that require retention are blocked under none; ZDR for those is per-account, per-model via AWS account manager (Claude ZDR eligibility managed by Anthropic). Enterprise Frontier Safeguards customers get ZDR for Claude Fable 5/5.1 through December 31, 2026.

Not covered

Models requiring retention (Claude Fable 5/5.1 require aws_review). CSAM-flagged image inputs/outputs may be stored and reviewed. Setting store=false alone does not guarantee ZDR.

From the docs

“none | Zero data retention. No request or response data is written to durable storage by AWS or shared with the model provider. On the Responses API, store defaults to false and store=true is rejected. Background mode is not available. Chat Completions and Messages requests are never retained.”

“If your organization requires zero data retention for compliance reasons and you need access to these models, contact your AWS account manager to discuss eligibility. ZDR access is evaluated on a per-account, per-model basis in coordination with the model provider.”

“ZDR eligibility for Claude models is managed by Anthropic. Contact your Anthropic account representative for support.”

“Customers that are eligible for the Enterprise Frontier Safeguards program will receive ZDR through December 31, 2026. After ZDR ends, all traffic will be retained for up to 30 days for automated offline abuse detection.”

Regional processing

US, EU and APAC

Geographic cross-Region profiles, or one AWS Region

Stated location

The AWS Region you call, unless you use cross-Region inference

Regions

• Single AWS Region (in-Region inference)
• Geographic cross-Region inference: US, EU, APAC (profile prefixes include us, eu, apac, in)
• Global cross-Region inference: any supported commercial Region

How to choose

Call the Bedrock endpoint in a chosen AWS Region, or pass a geographic or global inference profile ID as modelId. Data retention mode is set per Region.

What it covers

FAQ: content stored at rest in the Region where you use Bedrock. Geographic profiles keep processing within the geography; prompts/outputs may leave the source Region; abuse-detection retention is stored in the destination Region.

From the docs

“Any customer content processed by Amazon Bedrock is encrypted and stored at rest in the AWS Region where you are using Amazon Bedrock.”

“Cross-Region inference requests to an inference profile tied to a geography (such as US, EU, and APAC) stay within that geography. Your data remains in the AWS Regions where it originally resides. For example, a request made within the US is kept within the AWS Regions in the US. Your input prompts and output results might move outside of your source Region during cross-Region inference. To the extent we store data for abuse detection, your input prompts and output results will be stored in the destination region.”

“| Data residency | Within geographic boundaries (such as US, EU, and APAC) | Any supported AWS commercial Region worldwide | Choose Geographic for compliance requirements |”

“The setting is per-Region: it governs every request in the Region where you set it and does not propagate to other Regions.”

• Any AWS Region

  Copy

  bedrock.{region}.amazonaws.com

  bedrock-runtime / bedrock control plane (example shown: https://bedrock.us-east-1.amazonaws.com)

  Source: docs.aws.amazon.com

• Any AWS Region

  Copy

  bedrock-mantle.{region}.api.aws

  bedrock-mantle (example shown: https://bedrock-mantle.us-east-1.api.aws)

  Source: docs.aws.amazon.com

Hostnames taken from us-east-1 examples on the data retention page; the {region} pattern is inferred from those examples, not stated as a rule on that page.