AI vendors and your data

Vendors146Answers427

146 vendors · 427 answers

Fivetran Security Policy changed Fivetran, Oct 19, 2026

Fivetran · Oct 19, 2026 · 1 added, 13 removed · seen by the weekly check

The pageAll Fivetran terms
  1. Added: We use data minimization to limit: how often we access customer data for troubleshooting, direct access to customer data, and what we use when we interact with customer data, such as the use of obfuscated data rather than raw customer data values. We have tightly controlled troubleshooting environments with strict technical security safeguards, including: role-based access controls (SSO, reviews, etc.), logical separation, obfuscation, pseudonymization, short retention (obfuscated data is retained for 1 week), and secure storage.
  2. Removed: Fivetran cannot access your data without your approval.
  3. Removed: Depending on your deployment model, we may request access to your data in order to troubleshoot data integrity issues or fix broken connections or destinations. Learn how we troubleshoot for each deployment model type:
  4. Removed: When working on a support ticket, we may ask you to grant Fivetran access to your data for the next 21 days while we troubleshoot. You can allow or deny data access. If you grant us data access, you can revoke it at any time before the 21-day diagnostic period has expired.
  5. Removed: You can monitor your account's diagnostic data access in your Fivetran logs. View your account's access information using two log events, diagnostic_access_approved and diagnostic_access_granted .
  6. Removed: See our getting support documentation for more details.
  7. Removed: You can choose which level of data access to grant Fivetran. The levels of access are as follows:
  8. Removed: Source Access allows a Support Engineer or Connector Engineer to make API requests using the same token that Fivetran uses. We can then see the responses Fivetran receives from the source API during a sync and debug specific issues. With source access, we can also perform mock syncs, which use the same code as true syncs. While we can see each step of the sync to troubleshoot errors, it's not a real sync and so no data is written to the destination.
  9. Removed: Destination Access allows a Support Engineer or Connector Engineer to run SELECT queries using the same read-only permissions granted to the connection database user. We can then validate discrepancies and debug specific issues.
  10. Removed: Group Access grants source access for all connections within the given group, as well as destination access.
  11. Removed: Account Access grants source access for all connections within the account, as well as destination access for all groups.
  12. Removed: We create a unique debugging environment for each troubleshooting case. The debugging environment is created in the region of your data syncs, so it has the same data residency as your normal syncs. We automatically delete the debugging environment and all associated data after seven days.
  13. Removed: We follow multiple security protocols to protect your data while it's in the debugging environment. To connect to the debugging environment, our Support Engineer must authenticate with our company single-sign-on (SSO). The SSO requires multi-factor authentication (MFA), which enforces the use of a one-time passcode and a Fivetran-issued laptop. The laptop is managed by our IT department through mobile device management (MDM) software. Our MDM controls include the following:
  14. Removed: Since your connections run in your own environment, Fivetran doesn't have or request access to your data. We collaborate with you to troubleshoot, since only you have access to your data.
About this change
Page
fivetran.com/docs/security
Kind
Documentation
Text hash
0d617d62b16f to 531c1e328a98
Dated by
the weekly check that first read the new text