Cue Room Privacy Policy

The main sections below describe version 1.2: Apple sign-in, on-device script processing, and Pro-managed cloud AI. If you use version 1.1 (including build 53), its earlier personal OpenAI-key flow is described under If you use version 1.1 at the end of this page. The original PDF/photo upload and separate API billing disclosures still apply to version 1.1. Updating does not automatically convert an existing Plus subscription to Pro.

Effective October 16, 2026

Device processing

Scripts, recordings, import drafts, and rehearsal progress are stored on your device. PDFs and selected photos are decoded with native PDF text extraction or Vision OCR before analysis. Gemma is the primary on-device model on compatible devices after its one-time download. The model download starts automatically over Wi-Fi on first launch; verified existing downloads become active automatically. Built-in tools remain usable without it. Apple's on-device Foundation Models language model is the local fallback when Apple Intelligence is enabled and ready. Gemma runs while Cue Room is open. Local analysis and Apple voices do not send scripts to the developer or OpenAI. Microphone recognition requires on-device support and never silently switches to cloud recognition.

The automatic Gemma download uses about 2.6 GB, requires additional free storage during installation, and connects to Hugging Face and its model delivery hosts, which receive connection information such as your IP address. This download does not send scripts, recordings, or your Cue Room account credentials. Model files and temporary downloads are excluded from device backups. Downloads use Wi-Fi and can continue through iOS background transfers when Cue Room leaves the foreground. iOS controls background scheduling; force-quitting the app may interrupt a transfer. The download is recovered when Cue Room reopens; an explicit pause or removal remains paused. Progress and pause controls are available in the script library, and AI settings lets you resume or remove the download. Downloading Gemma does not enable cloud processing.

Pro AI processing

Cloud AI is included with an active, server-verified Cue Room Pro subscription. Signing in alone does not upload scripts. Pro activation presents the third-party AI sharing disclosure and asks for permission once for the signed-in account. Cue Room then checks Pro access automatically; there is no separate cloud-processing mode in Settings. A previous refusal is preserved until you explicitly activate Pro AI. Without a verified Pro entitlement, new cloud requests are disabled. Existing paid Plus subscriptions use on-device processing; the future Free tier has no generated voices, including Apple voices.

Import tries local analysis first. If it fails or is unavailable, activated Pro AI can send recognized script text and bounded role context through the Cue Room backend to OpenAI for annotation. Original imported PDFs/photos remain on device. Saving an imported Pro script queues contextual voice casting: bounded play excerpts, role notes, and sample dialogue help select synthetic voices. Dialogue and the selected voice are then sent for speech generation. Confirmed audio is saved on this device and reused for unchanged text, voice, model, and audio format. Acting directions do not change the current cloud voice delivery. Recorded-take transcription uploads the take you select, including your voice and any spoken personal information. Paraphrased cue checks send recognized words and your expected line as text, without microphone audio.

Our backend and OpenAI receive any personal information contained in that content. Our backend hosting receives connection information including your IP address; OpenAI receives requests from the backend. The backend holds the OpenAI credential; the app never receives it. Generated voices are synthetic and do not clone performers.

Accounts, subscriptions, and safeguards

Sign in with Apple is required before entering Cue Room, including the sample rehearsal; the launch offering requires an active paid subscription (the implemented Free tier is not offered at launch). Saved-data management and the bundled privacy and terms documents remain available before sign-in. The app saves its session securely in the device Keychain and renews it automatically; a revoked or expired authorization may require signing in again. A saved account and locally verified subscription allow on-device tools during temporary service outages. Apple sign-in identifies your Cue Room account without requesting your name or email. The backend verifies Apple identity tokens and uses encrypted Apple refresh credentials to manage account authorization. App Store signed subscription transactions, transaction/account identifiers, validity, and revocation information automatically verify cloud access after sign-in. Payment details are handled by Apple and are not sent to OpenAI.

The backend stores hashed session credentials, account status, and bounded usage counters to enforce request, concurrency, and daily and dollar allowances across apps. The current allowance is up to $5 USD of provider processing per Cue Room account per UTC calendar month, subject to the shared service budget and availability. These are service usage limits, not additional charges to your Apple Account. Script and audio request bodies are processed for the request and are not saved to the backend database or application logs. Render and Neon, our hosting and database providers, process the account and operational data needed to run this service.

Controls and deletion

In AI settings, Withdraw Pro AI permission stops new cloud requests and cancels queued voice preparation. Already-sent requests may finish. Withdrawal preserves your signed-in account, subscription, local tools, and saved audio; it does not delete content already held by a provider. The choice remains in effect for this account until you explicitly complete Activate Pro AI again.

You can sign out, clear downloaded audio, or delete your Cue Room account from AI settings, including without an active subscription. Deletion revokes Apple authorization and removes active session/account access. Opaque identity, transaction binding, and quota tombstones may be retained for abuse prevention and to prevent deletion from resetting usage allowances. These limited records remain after account deletion; contact support with requests about them. Deleting the Cue Room account does not delete local scripts, exported copies, or OpenAI's provider records, and does not cancel an App Store subscription. Use App Store account settings to cancel it.

OpenAI handles submitted content according to its API terms and data controls. Account/provider retention can differ from local deletion. Review OpenAI data controls for current handling. The backend requests text responses with stored response state disabled.

Other permissions and sharing

Camera, photo selection, microphone, local network, and nearby cast sharing remain user controlled. Exports and nearby sharing send the selected data to the recipients you choose. Device backups and exported copies are managed separately.

Contact

Contact curtaincallsupport@icloud.com for support and privacy requests.

If you use version 1.1

Version 1.1 does not have a Cue Room developer account or shared backend. Scripts, recordings, drafts, and progress are stored locally. Its document importer uses OpenAI rather than local OCR or local AI: after you enable OpenAI requests and choose Continue with OpenAI, selected PDF pages, photos/images, or decoded Word/text content are sent directly to OpenAI. Uncached generated voices send dialogue, voice choice, acting directions, bounded role context, and the preceding line. Recorded-take transcription sends your selected audio; optional paraphrased cue checks send recognized words and the expected line without microphone audio. These requests can contain personal information and expose connection information such as your IP address to OpenAI. Requests use your own API key and provider account; Cue Room Plus does not include API credits. Saving a key alone does not send content.

The version 1.1 key is stored in device-only Keychain, separate from exported scripts and nearby transfers. Apple & OpenAI settings lets you disable new requests, remove the key, or clear downloaded AI audio. Saved scripts can use installed Apple voices and supported on-device recognition without a key. Temporary import attachments and completed sections persist locally for recovery until you save or discard the draft. Generated audio is cached locally; deleting a script does not clear that shared cache. Manage saved data remains available without paid access. Clearing local data cannot delete exported copies, backups, or provider-held records. Deleting the app does not cancel a subscription; remove the saved key in the app first if you want it removed from Keychain.

OpenAI handles version 1.1 requests under its API terms and your account settings. Text analysis and cue checks disable stored response state; provider abuse-monitoring or other retention may still apply. See OpenAI data controls and OpenAI privacy policy. Cue Room cannot delete provider-held records. Apple handles billing; version 1.1 checks StoreKit locally without sending purchase records to a Cue Room server. User-initiated exports/nearby sharing send chosen scripts and recordings to recipients who may retain copies. If you email support, we receive your address and chosen message/attachments, use them to respond, and retain them as needed for support and applicable obligations; contact curtaincallsupport@icloud.com about access, correction, or deletion.