LiveKit’s data processing agreement changed.

− [livekit](https://github.com/livekit)/agents13.9K
− - [agents13.9K](https://github.com/livekit/agents)
− - [livekit20.6K](https://github.com/livekit/livekit)
− Last Updated: 07/22/2026
− "Services" means the products and services provided by LiveKit or its Affiliates, as applicable, that are (a) used by Customer, including, without limitation, products and services that are on a trial basis or otherwise free of charge or (b) ordered by Customer under an Order Form. Services include products and services that provide both (x) platform services, including access to any application programming interface ("*LiveKit API*") and (y) where applicable, communications services used in connection with the LiveKit APIs.
− b. If Customer is a Processor on behalf of a Third-Party Controller, then Customer: is the single point of contact for Company; must obtain all necessary authorizations from such Third-Party Controller; and undertakes to issue all instructions and exercise all rights on behalf of such other Third-Party Controller.
− 2.3 LiveKit as a Controller. Customer acknowledges that LiveKit may process Personal Data as a Controller under Data Protection Laws, including in relation to Customer Account Data and Operational Metrics (including limited phone number metadata). Customer acknowledges that LiveKit will process Observability Data as a Controller where Customer elects to join the Model Improvement Program, or where Customer makes use of a version of the Services for which no Fees are payable. LiveKit will process Customer Account Data and Operational Metrics as a Controller for purposes which include (a) managing the relationship with Customer; (b) carrying out LiveKit's core business operations, such as accounting and filing taxes; (c) detecting, preventing, or investigating security incidents, fraud, and other abuse or misuse of the Services; (d) providing, optimizing, and maintaining the Services, platform, and security; (e) performing identity verification; and (f) complying with applicable law or regulation or as otherwise permitted under Applicable Data Protection Law and otherwise in accordance with the LiveKit Privacy Policy.
− 2.5 Region Pinning. Where Customer configures the Services to use a specific geographic region (a "*Pinned Region*") through the LiveKit Cloud dashboard or API, LiveKit will process the following categories of Customer Content in the Pinned Region:
− -
− Developer Application Data, to the extent necessary to host, execute, and maintain Customer's deployed applications or hosted agents;
− -
− User Content, including audio, video, text, and session metadata transmitted through the Services, for purposes of real-time media transport, signaling, and delivery within the Pinned Region; and
− Inference Data, solely where Customer selects an Inference Provider that operates infrastructure within the same Pinned Region.
− Region pinning limits the routing and processing of the above data categories to compute and network resources located within the designated Pinned Region. Region pinning does not provide cross-region failover; if the selected Pinned Region becomes unavailable, Services in that region may be disrupted until restored.
− Customer acknowledges that certain other categories of Personal Data are not subject to region pinning:
− Observability Data, telemetry, and related logs are stored and processed in the United States regardless of the selected Pinned Region;
− where Customer selects an Inference Provider that operates outside the Pinned Region, the corresponding Inference Data will be transmitted to and processed by that provider in accordance with Customer's configuration, the Inference Provider's usage policies, and LiveKit's zero data retention, including Inference Providers contractually agreeing to not retain, log, train, fine tune or otherwise use Inference Data for model improvement ("*ZDR*");
− beta, preview, or experimental features are provided "as is" and may not support region pinning. LiveKit makes no representation or warranty that such features will adhere to Customer's Pinned Region configuration.
− 6.1 Authorization for Onward Sub-processing. Customer provides a general authorization for LiveKit to engage Sub-processors. A list of LiveKit's current Sub-processors is in [https://livekit.com/legal/sub-processors](https://livekit.com/legal/sub-processors). LiveKit agrees to impose contractual data protection obligations on any sub-processor it appoints that require such Sub-processor to protect Customer Content as required by Applicable Data Protection Law. LiveKit will remain liable for any breach of this Addendum that is caused by an act, error, or omission of its Sub-processors.
− 9. Return or Deletion of Customer Content. LiveKit will, in accordance with Section 3 (Duration of the Processing) of Schedule 1 (Details of Processing) of this Addendum, delete or return to Customer any Customer Content stored within the Services. LiveKit may maintain encrypted backups of Customer Content for up to 30 days following deletion for disaster-recovery purposes, after which such backups are securely purged.
− 10.1 Security Measures. LiveKit has implemented and will maintain the technical and organizational security measures as set forth in the LiveKit Security Overview document available at [https://livekit.com/security/overview](https://livekit.com/security/overview). Additional information about LiveKit's technical and organizational security measures to protect Customer Content is set forth in Schedule 2 (Technical and Organizational Security Measures) of this Addendum.
− Each party's and all of its Affiliates' liability, taken together in the aggregate, arising out of or relating to this Addendum, the EU Standard Contractual Clauses, and any other data protection agreements in connection with the Agreement (if any), shall be subject to any aggregate limitations on liability set out in the Agreement. Without limiting the parties' obligations under the Agreement, each party agrees that any regulatory penalties incurred by one party (the "*Incurring Party*") in relation to the Customer Content that arise as a result of, or in connection with, the other party's failure to comply with its obligations under this Addendum or any Applicable Data Protection Law shall count toward and reduce the Incurring Party's liability under the Agreement as if it were liability to the other party under the Agreement.
− Neither party will be responsible for any fines issued or levied under Applicable Data Protection Law against the other party by a regulatory authority or governmental body in connection with such other party's violation of the Applicable Data Protection Law.
− Modules 2 and 3 | Inference Data | Generate outputs from third-party inference/model providers. | Transmission of prompts/inputs and return of outputs; no logging or retention by LiveKit or Inference Providers used through LiveKit Inference, except during the transitory period necessary to complete the applicable inference request. | Ephemeral; not retained beyond the duration necessary to complete the applicable inference request. | Customer personnel and end users. |
− Modules 1, 2 and 3 | Observability Data | Provide monitoring, debugging, analytics, and performance insights. De-identified and aggregated observability data from free-tier accounts or customers who explicitly opt in to the Model Improvement Program may be used to enhance AI-powered features and model accuracy. | Collection of logs, traces, transcripts, error/latency metrics; session-level telemetry. | Retained up to 30 days by default from the date of collection unless Customer requests earlier deletion; with encrypted backups maintained for no more than an additional 30 days before deletion. | Customer personnel and end users. |
− Module 1 | Customer Account Data | Manage billing, account administration, support, compliance. | Collection of account details, billing, support logs, credentials. | Retained for the life of the account + up to 30 days after closure; longer where required for tax, audit, or compliance with applicable laws. | Customer personnel. |
− Measures for ensuring data minimization | LiveKit does not collect or store any Personal Data. Data for each session's participants is anonymized using randomly-generated identifiers. Customer may choose to assign a separate identifier for a session participant, however LiveKit uses the generated identifier to relate all data. |
− Measures for ensuring limited data retention | Telemetry data is automatically purged after its configured retention period. Inference Data is not logged or retained by LiveKit or Inference Providers used through LiveKit Inference beyond the duration necessary to complete the applicable inference request, and is not used for training, fine-tuning, or model improvement. |
− The description of the transfer is set forth in (Nature and Purpose of the Processing) of Schedule 1 (Nature and Purpose of the Processing), (Details of the Processing).
+ [livekit](https://github.com/livekit)/agents14.0K
+ - [agents14.0K](https://github.com/livekit/agents)
+ - [livekit20.7K](https://github.com/livekit/livekit)
+ Last Updated: 09/04/2026
+ "Inference Provider" means a third-party provider of speech-to-text, text-to-speech, language, or other machine-learning model services that LiveKit engages as a Sub-processor to process Inference Data in connection with LiveKit Inference.
+ "LiveKit Inference" means the LiveKit-managed gateway through which Customer may access models operated by Inference Providers.
+ "Services" means the products and services provided by LiveKit or its Affiliates, as applicable, that are (a) used by Customer, including, without limitation, products and services that are on a trial basis or otherwise free of charge or (b) ordered by Customer under an Order Form. Services include products and services that provide both (x) platform services, including access to any application programming interface ("*LiveKit API*") and LiveKit Inference, and (y) where applicable, communications services used in connection with the LiveKit APIs.
+ b. If Customer is a Processor on behalf of a Third-Party Controller, then Customer: is the single point of contact for LiveKit; must obtain all necessary authorizations from such Third-Party Controller; and undertakes to issue all instructions and exercise all rights on behalf of such other Third-Party Controller.
+ 2.3 LiveKit as a Controller. Customer acknowledges that LiveKit may process Personal Data as a Controller under Data Protection Laws, including in relation to Customer Account Data and Operational Metrics (including limited phone number metadata). LiveKit will process Customer Account Data and Operational Metrics as a Controller for purposes which include (a) managing the relationship with Customer; (b) carrying out LiveKit's core business operations, such as accounting and filing taxes; (c) detecting, preventing, or investigating security incidents, fraud, and other abuse or misuse of the Services; (d) providing, optimizing, and maintaining the Services, platform, and security; (e) performing identity verification; (f) sending Customer marketing and promotional communications; and (g) complying with applicable law or regulation or as otherwise permitted under Applicable Data Protection Law and otherwise in accordance with the LiveKit Privacy Policy.
+ 2.5 Region Pinning. Where, at Customer's request submitted to LiveKit support, LiveKit configures the Services to route Customer's real-time media and signaling through a specific geographic region (a "*Pinned Region*"), LiveKit will process User Content, including audio, video, text, and session metadata transmitted through the Services, in the Pinned Region for purposes of real-time media transport, signaling, and delivery within the Pinned Region.
+ Region pinning limits the processing of User Content to compute and network resources located within the designated Pinned Region. Region pinning does not provide cross-region failover; if the selected Pinned Region becomes unavailable, Services in that region may be disrupted until restored.
+ Customer acknowledges that the following categories of data are not subject to region pinning and are processed in locations determined by other Customer configuration settings or by LiveKit's infrastructure, as follows:
+ Developer Application Data is processed in the region in which Customer elects to deploy its applications or hosted agents, which is configured independently of, and may differ from, the Pinned Region;
+ Observability Data, telemetry, and related logs are stored and processed in the United States or Europe, as determined by the data region selected for Customer's project, and not by the Pinned Region;
+ Inference Data is transmitted to and processed by the Inference Provider selected by Customer in the region in which that provider operates, as determined by Customer's inference configuration (including, where LiveKit makes it available, an optional inference region setting for hosted agents) and not by the Pinned Region. Inference Data remains subject to ZDR as set forth in Section 2.6 (Zero Data Retention for LiveKit Inference) regardless of the region in which it is processed;
+ Beta, preview, or experimental features are provided "as is" and may not support region pinning. LiveKit makes no representation or warranty that such features will adhere to Customer's Pinned Region.
+ 2.6 Zero Data Retention for LiveKit Inference. LiveKit maintains a zero data retention posture for Inference Data processed through LiveKit Inference ("*ZDR*"). Under ZDR, neither LiveKit nor any Inference Provider will (a) retain Inference Data beyond the period necessary to complete the applicable inference request; (b) log Customer inputs or outputs, excluding non-content security and operational metadata such as request counts, latency metrics, and error events that do not include Inference Data; or (c) use Inference Data to train, fine-tune, or otherwise improve any model or service.
+ LiveKit enforces ZDR through a combination of (i) contractual requirements with Inference Providers and (ii) technical controls at the gateway layer, including request-level configuration and production verification. ZDR applies by default to all Inference Data processed through LiveKit Inference and is not conditioned on Customer's subscription tier, the selection of a Pinned Region, or any Customer configuration setting.
+ ZDR does not apply to: (1) Observability Data, which is processed and retained in accordance with Schedule 1 (Details of Processing); (2) voice sample recordings and derived voice models submitted or generated where Customer or its end users elect to use voice reproduction or voice cloning features, which are retained in accordance with Schedule 1; and (3) Customer Content that Customer elects to record, log, store, or export through features made available by the Services.
+ Where LiveKit makes available features intended to detect, mask, or redact Personal Data within User Content or Inference Data, such features are provided on a best-efforts basis. LiveKit does not represent or warrant that any such feature will identify or remove all Personal Data, and Customer remains responsible for determining whether the Services are suitable for its intended use.
+ 6.1 Authorization for Onward Sub-processing. Customer provides a general authorization for LiveKit to engage Sub-processors. A list of LiveKit's current Sub-processors is at [https://livekit.com/legal/sub-processors](https://livekit.com/legal/sub-processors), which identifies which Sub-processors are Inference Providers. LiveKit agrees to impose contractual data protection obligations on any sub-processor it appoints that require such Sub-processor to protect Customer Content as required by Applicable Data Protection Law. LiveKit will remain liable for any breach of this Addendum that is caused by an act, error, or omission of its Sub-processors.
+ 9. Return or Deletion of Customer Content. LiveKit will, in accordance with the column marked "Duration / Retention" in Schedule 1 (Details of Processing) of this Addendum, delete or return to Customer any Customer Content stored within the Services. LiveKit may maintain encrypted backups of Customer Content for up to 30 days following deletion for disaster-recovery purposes, after which such backups are securely purged.
+ 10.1 Security Measures. LiveKit has implemented and will maintain the technical and organizational security measures as set forth in the LiveKit Security Overview document available at [https://livekit.com/security](https://livekit.com/security). Additional information about LiveKit's technical and organizational security measures to protect Customer Content is set forth in Schedule 2 (Technical and Organizational Security Measures) of this Addendum.
+ (a) Each party's and all of its Affiliates' liability, taken together in the aggregate, arising out of or relating to this Addendum, the EU Standard Contractual Clauses, and any other data protection agreements in connection with the Agreement (if any), shall be subject to any aggregate limitations on liability set out in the Agreement. Without limiting the parties' obligations under the Agreement, each party agrees that any regulatory penalties incurred by one party (the "*Incurring Party*") in relation to the Customer Content that arise as a result of, or in connection with, the other party's failure to comply with its obligations under this Addendum or any Applicable Data Protection Law shall count toward and reduce the Incurring Party's liability under the Agreement as if it were liability to the other party under the Agreement.
+ (b) Neither party will be responsible for any fines issued or levied under Applicable Data Protection Law against the other party by a regulatory authority or governmental body in connection with such other party's violation of the Applicable Data Protection Law.
+ Modules 2 and 3 | Inference Data | Generate outputs from third-party inference/model providers. | Transmission of prompts/inputs and return of outputs; no logging or retention of Inference Data by LiveKit or Inference Providers except during the transitory period necessary to complete the applicable inference request. Non-content security and operational metadata that does not include Inference Data (such as request counts, latency metrics, and error events) may be generated and retained for billing, security, and service-reliability purposes. | Ephemeral; Inference Data is not retained beyond the duration necessary to complete the applicable inference request. Non-content operational metadata described above is retained as Operational Metrics. | Customer personnel and end users. |
+ Modules 2 and 3 | Voice Sample Data | Enable voice reproduction and voice cloning features where Customer or its end users elect to use them. | Storage of submitted audio samples; generation and storage of derived voice models; synthesis from stored voice models. | Retained until deleted by Customer or its end user, and in any event no longer than twelve (12) months following the last use of the associated voice. Not subject to ZDR. | Customer personnel and end users. |
+ Modules 2 and 3 | Observability Data | Provide monitoring, debugging, analytics, and performance insights. | Collection of logs, traces, transcripts, error/latency metrics; session-level telemetry. | Retained up to 30 days by default from the date of collection unless Customer requests earlier deletion; with encrypted backups maintained for no more than an additional 30 days before deletion. | Customer personnel and end users. |
+ Module 1 | Customer Account Data | Manage billing, account administration, support, compliance, sending marketing communications from LiveKit. | Collection of account details, billing, support logs, credentials. | Retained for the life of the account + up to 30 days after closure; longer where required for tax, audit, or compliance with applicable laws. | Customer personnel. |
+ Measures for ensuring data minimization | LiveKit limits the Personal Data it collects and stores to what is necessary to provide the Services, as set forth in Schedule 1. Data for each session's participants is pseudonymized using randomly-generated identifiers. Customer may choose to assign a separate identifier for a session participant, however LiveKit uses the generated identifier to relate all data. |
+ Measures for ensuring limited data retention | Telemetry data is purged in accordance with the retention periods set forth in Schedule 1. Inference Data is not logged or retained by LiveKit or Inference Providers beyond the duration necessary to complete the applicable inference request, and is not used for training, fine-tuning, or model improvement. Non-content operational metadata that does not include Inference Data is retained as Operational Metrics. |
+ The description of the transfer is set forth in Schedule 1 (Details of the Processing).
+ ## Updates[#](#updates)
+ As our business grows and evolves, the data processing addendum may also change. We will seek to provide the owners of customer accounts with notice of changes to our data processing addendum to the extent required under the Terms. If you would like to be notified of updates to our data processing addendum, please fill out this [form](https://41rcwn.share-na2.hsforms.com/28yBQYbkFRCepixb4WU35zQ).