Claude on AWS: data residency & retention

This is Anthropic's own Claude API served on AWS infrastructure (Anthropic PBC is the operating entity; Anthropic Ireland, Limited contracts with EEA/Swiss/UK customers). Anthropic's privacy policy states personal data is transferred to its servers in the US, but Anthropic does not publish a list of AWS regions/datacenters used for its first-party API.

No self-serve region pinning is documented for Anthropic's first-party API; regional deployment of Claude is generally achieved via cloud marketplaces (e.g. Bedrock/Vertex) rather than this endpoint.

Standard API inputs and outputs are automatically deleted from Anthropic's backend within 30 days of receipt or generation (exceptions for trust & safety flags and legal holds). Prompts/outputs for designated 'covered models' (Mythos-class) are retained 30 days for safety purposes on every platform.

Commercial Terms: 'Anthropic may not train models on Customer Content from Services.' (The consumer privacy policy's train-unless-you-opt-out language applies to consumer products, not the commercial API.)

Zero-data-retention agreements are available for API customers but are not automatic — organizations must apply/qualify; under ZDR, data is not stored at rest after the response except as required by law or to combat misuse.