Backfrom AkashML’s own policies · verified 2026-09-02
AkashML: data residency & retention
Headquarters
🇺🇸 United States (US)
Where inference runs
Not disclosed
Disclosed architecture, not geography: 'the Services are hosted on the Akash Network, a decentralized compute network operated by third-party node operators.' Node locations are inherently variable and no geographic regions are published, so no bloc can be asserted. Providers are 'contractually prohibited from retaining or repurposing' processed data.
Region pinning
No
No region-pinning controls disclosed; on a decentralized third-party node network there is no published residency guarantee.
Prompt retention
Not retained
Privacy policy: 'we do not retain prompts or transient inputs after the request has been processed.' Exceptions: failed-request inputs 'may be logged briefly' for troubleshooting then deleted per retention schedule; outputs a user affirmatively saves to their account are kept until deleted.
Trains on API data
No
Explicit: 'We do not use... prompts, generated Outputs, or other User Content to train, fine-tune, or improve any AI model' (applies to proprietary and third-party models).
Zero data retention
Available
Not marketed as a formal ZDR program, but the default posture per the privacy policy is zero prompt retention after processing.
Compiled from AkashML’s published privacy policy, terms, and documentation on 2026-09-02. “Not disclosed” means AkashML publishes nothing on the point. Policies change without notice, so verify against the sources above before relying on this for compliance decisions.