Zero Data Retention
Zero Data Retention
Stop Inworld from retaining your request and response content, set for your whole organization or per workspace, across TTS, STT, LLMs, and the Realtime API
Customers in healthcare, finance, and other regulated industries can ensure that eligible customer inputs and generated outputs are not persistently stored in our systems after processing completes. Zero Data Retention is available across Inworld TTS, STT, hosted LLMs, and the Realtime API. When enabled, eligible requests are processed in memory and their inputs and outputs are immediately redacted from our logging systems once processing completes. What is covered, and the exceptions that remain, varies by product — see the sections below for details.
Turn it on
Zero Data Retention is available on the Enterprise and Enterprise - Trial plans and is configured by Owners and Admins on the Data Retention tab of the Team page.
There are two levels, and you can use either or both:
- Organization default — applies to every workspace in your organization. Set it on the Data Retention tab.
- Workspace override — turns Zero Data Retention on or off for one workspace, whatever the organization default is. Set it in that workspace's own settings.
A workspace's setting wins where it has one. Where it does not, the workspace follows the organization default:
| Organization default | Workspace override | Result for that workspace |
|---|---|---|
| On | (inherit) | On |
| On | Off | Off |
| Off | On | On |
| Off | (inherit) | Off |
Changes take effect on the next request. Requests already in flight finish under the setting they started with.
Turning Zero Data Retention on stops us retaining content from that point forward. It does not delete content already collected before you turned it on. If you need existing data removed, contact support.
If you are not on an Enterprise plan
The plan requirement applies in both directions. On any other plan you can neither turn Zero Data Retention on nor turn it off yourself — the controls show your real setting but are disabled. Organizations and workspaces that already have it on keep it.
This matters if your plan changes: an organization that had Zero Data Retention on before moving off Enterprise keeps it, and turning it off from there is a support request rather than a self-serve change. Contact sales about Enterprise, or support to change a setting you can no longer reach.
Inworld may also enable Zero Data Retention for a workspace independently of this setting, for example while a support arrangement or contractual commitment is in effect. Where that applies, content is not retained even if the setting here reads as off.
Moving a workspace between organizations
A workspace that inherits its setting follows the organization it moves to. If the destination organization has Zero Data Retention on, the workspace gets it on arrival; if the destination has it off, an inheriting workspace loses it. A workspace with its own override keeps that override across the move.
TTS
When enabled, all customer text inputs sent to our text-to-speech engine are processed in memory to generate audio. Once complete, both the text and the generated audio are immediately redacted from our logging systems. This applies to all text sent for speech synthesis, including requests using cloned voices.
Asynchronous synthesis jobs work differently from streaming requests. Because results are retrieved after the job completes rather than returned on the same connection, the generated audio and timestamp artifacts are stored in Inworld-managed storage so you can download them, and are automatically deleted after 7 days. Result artifacts are reachable only through expiring signed URLs. Input text redaction applies to asynchronous jobs the same way it applies to streaming requests.
| Product | Type | Eligible for Zero Data Retention? |
|---|---|---|
| Text-to-Speech | Text Input | Yes |
| Text-to-Speech | Audio Output | Yes |
| Text-to-Speech (async jobs) | Text Input | Yes |
| Text-to-Speech (async jobs) | Result artifacts (audio, timestamps) | Retained up to 7 days for delivery, then auto-deleted |
| Text-to-Speech | Usage metadata (model, character count, timestamps, request and API key identifiers) | No (retained for billing) |
| Voice Cloning | Audio Samples | No |
| Voice Design | Voice Description / Preview Script | No |
STT
When a request is processed, customer audio is transcribed in memory. Once complete, both the original audio and the transcription output are immediately redacted from our logging systems. This applies to requests, both sync (file upload) and WebSocket (streaming), processed through the Inworld STT API.
| Data | Type | Eligible for Zero Data Retention? |
|---|---|---|
| Audio Input | Customer-submitted audio (file or stream) | Yes |
| Transcription Output | Transcribed text returned by the API | Yes |
| Usage Metadata | Model, audio duration, timestamps, request and API key identifiers | No (retained for billing) |
Zero Data Retention applies to all STT models available through the Inworld STT API:
| Model | Details |
|---|---|
inworld/inworld-stt-1 | Processed in memory; redacted from logs after completion |
LLM Router
Zero Data Retention is currently available only for Inworld hosted LLM models. Requests routed to third-party model providers are not covered.
When enabled, all customer prompts sent to Inworld hosted LLMs are processed in memory to generate a response. Once complete, neither the prompt nor the generated output is written to our logging or storage systems. This applies to all content sent for completion, including system prompts and tool interactions.
| Product | Type | Eligible for Zero Data Retention? |
|---|---|---|
| Inworld hosted LLM | Prompt / Input messages | Yes |
| Inworld hosted LLM | Model output / Completion | Yes |
| Inworld hosted LLM | Usage metadata (model, token counts, timestamps, request and API key identifiers) | No (retained for billing) |
| Third-party model providers | Prompt & Output | No |
Realtime API
When enabled, the content of a Realtime session — prompts, transcripts, and streamed response text — is not written to our logging systems, and request and response payloads are excluded from diagnostic traces.
A Realtime session takes its setting when the connection is established, so a change made mid-session applies from the next connection rather than to the one already open.
| Product | Type | Eligible for Zero Data Retention? |
|---|---|---|
| Realtime API | Prompts and instructions | Yes |
| Realtime API | Transcripts and response text | Yes |
| Realtime API | Usage metadata (models, token counts, audio duration, character count, timestamps) | No (retained for billing and service monitoring) |
Enterprise-level security
Inworld's platform is SOC 2 Type II certified and HIPAA and GDPR compliant. For full details on our security posture, certifications, and data handling practices, see our Security page.
Organization or workspace
Set Zero Data Retention once for your whole organization, or per workspace where a single team has stricter requirements than the rest.
Production Scale
Whether you're handling protected health information or sensitive financial data, you get the security guarantees needed for production deployment at scale.
FAQs
Who can turn Zero Data Retention on?
Owners and Admins of the organization, on the Data Retention tab of the Team page. It is available on the Enterprise and Enterprise - Trial plans. On other plans the setting is visible but not editable, in either direction.
Can we enable Zero Data Retention for only specific workspaces?
Yes. Leave the organization default off and turn it on for the workspaces that need it, in each workspace's own settings. The reverse also works: turn it on for the organization and switch off the few workspaces that do not need it.
What happens to data collected before we turned it on?
It is unaffected. Zero Data Retention stops content being retained from the moment it takes effect; it does not remove content already stored. Contact support if you need existing data deleted.
How does Zero Data Retention affect debugging?
With Zero Data Retention enabled, original inputs and outputs cannot be retrieved. Debugging and troubleshooting capabilities are limited to non-sensitive metadata only.
How does Inworld TTS Zero Data Retention work?
TTS Zero Data Retention protects privacy by redacting all input texts from logs, including text sent to Cloned Voices, while only retaining the initial source audio required to build the voice itself.
Does Zero Data Retention cover asynchronous synthesis?
Yes, with one bounded exception. Input text and generated audio are never written to our logging systems. But an asynchronous job's results must exist somewhere for you to download them, so the generated audio and timestamp artifacts are stored in Inworld-managed storage until you retrieve them. These artifacts are reachable only through expiring signed URLs and are automatically deleted after 7 days. Turning Zero Data Retention on does not remove artifacts already inside that window.
Does TTS Zero Data Retention affect audio output quality?
No. The TTS engine still receives the text to generate the audio. The redaction happens only at the logging layer, ensuring the record of what was said is not saved.
Does Zero Data Retention cover Voice Design?
Not for the Voice Design creation step itself. The voice description and script you provide in Voice Design are inputs used to build your voice, similar to the source audio in Voice Cloning. These inputs are retained in order to create and support the voice.
Once your Voice Design voice is published, TTS Zero Data Retention applies when enabled. Any text you synthesize with that voice is treated like any other TTS request.
How does Inworld STT Zero Data Retention work?
STT Zero Data Retention protects privacy by redacting all customer audio and transcription outputs from logs immediately after processing. Only non-sensitive usage metadata (request duration, model ID, timestamps) is retained for billing and service monitoring.
Does Zero Data Retention apply to all STT models?
Yes. All models available through the Inworld STT API operate under Zero Data Retention. Each upstream provider's specific guarantees are listed in the supported models table above.
Is my audio used for model training?
No. Audio processed through the Inworld STT API is never used for training by Inworld or any upstream provider.
Does STT Zero Data Retention affect transcription quality?
No. The STT engine still receives the audio to generate the transcript. The redaction happens only at the logging layer, ensuring no record of the audio or transcript is saved.
How does Inworld LLM Zero Data Retention work?
LLM Zero Data Retention protects privacy by never writing prompts or model outputs to our logs. It is currently available for Inworld hosted LLM models.
Which LLM models support Zero Data Retention?
Zero Data Retention is currently available only for Inworld hosted LLM models. Requests routed to third-party model providers are not covered.
Does LLM Zero Data Retention affect output quality?
No. The model still receives the full prompt to generate its response. Zero Data Retention applies only at the logging layer: the prompt and output are never written there, so no record of the conversation is saved.