Privacy Policy

Last updated: 2026-09-07

SceneStealer (“we,” “us”) provides software that helps live-performance operators turn full-length show recordings into short highlight clips for social media. This policy explains what data we collect, why, and how you can control it.

Who this applies to

SceneStealer is a business-to-business product. Our customers (“tenants”) are theater companies, dance companies, concert venues, and similar live-performance organizations; each tenant’s own staff are the people who sign in and use the product. This policy covers both tenants and the individual people who use a tenant’s account.

What we collect

• Account information: name, email address, and organization membership, handled by our authentication provider, Clerk.
• Content you upload: full-length show recordings (video/audio files) you upload for processing.
• Content we generate from it: automatic transcripts, scene-boundary and audio-energy analysis, AI-scored highlight suggestions, and the rendered clips produced once you accept a suggestion.
• File metadata: when present in an uploaded recording’s own file metadata, we read the recording date, device model, and location. Location is never stored or shown as exact coordinates or a street address — only a city name, and a venue name when the location matches a mapped business. These are offered as optional fields for your own caption templates.
• Publishing connections: if you choose to connect a social/video platform (currently YouTube; Instagram and Facebook once available), we store the access token that platform issues to us, scoped to the permissions you granted.
• Billing information: handled directly by Stripe; we do not store your card details ourselves.

How we use it

We use your uploaded recordings solely to generate clip suggestions for your review. Suggested clips are never published automatically — a person on your team must explicitly accept a clip, and separately, explicitly trigger a render and a publish. We do not use your content to train AI models, and we do not sell your data. See AI Transparency below for exactly what our AI-assisted steps do and don’t see.

AI Transparency

This section explains, specifically, what our AI-assisted steps do, what they see, and who stays in control of what actually gets published.

What AI actually does. Two AI-assisted steps run automatically when you upload a recording and start analysis: transcription (your recording’s audio track is transcribed to text, with timestamps), and highlight suggestion (an AI model reads the transcript, a list of audio-energy spikes — moments louder than the surrounding baseline, typically applause, laughter, or audience reaction — and the real scene/cut boundaries our own, non-AI, scene-detection step already found, then proposes a ranked list of candidate clip windows with a short written reason for each). That’s the full extent of it: there is no AI step that edits your footage, alters what was said or done on stage, or generates any synthetic video, audio, or narration. A suggested clip is always a real, unmodified excerpt of your original recording — the AI only picks the start and end points.

What the AI model sees — and what it doesn’t. The highlight-suggestion model never receives your raw video or audio. It only receives text: the transcript, the timestamps of audio-energy spikes, and the timestamps of real scene boundaries. Your actual video and audio files are never uploaded to, or processed by, the AI model that suggests highlights.

A person decides what actually happens. Nothing the AI suggests is published, or even rendered, on its own. Every suggested clip sits as a proposal until someone on your team accepts it (or draws their own clip manually, which never involves the AI at all) — the AI has no part in triggering a render. Publishing or scheduling a rendered clip to a connected platform is a separate, explicit action, taken by a person, every time. You can also reject any suggestion outright, and adjust a clip’s start/end points before accepting it, whether or not it came from an AI suggestion.

Which AI providers are involved. Transcription runs on Groq’s hosted Whisper model. Highlight suggestion runs on Anthropic’s Claude — see Third-party service providers below for every provider we use.

Disclosing AI involvement on platforms you publish to. Some social platforms have their own rules about disclosing AI-assisted content. Because the clips SceneStealer produces are always real, unaltered excerpts of your own recording — the AI only chooses which moment to excerpt — most of these platform rules (generally aimed at synthetic or substantially altered media) are unlikely to apply. We’re not able to give platform-specific legal advice, though, so if you’re unsure, check the current disclosure policy of whichever platform you’re publishing to.

Third-party service providers

We rely on the following providers to run SceneStealer. Each only receives the data it needs to perform its specific function:

• Clerk — authentication and organization membership.
• Neon — our Postgres database (tenant, video, and clip records).
• Cloudflare (R2, Workers, Pages) — file storage and application hosting.
• Fly.io — runs the video-processing jobs (transcription, scene detection, rendering).
• Groq — transcribes your recording’s audio track (Whisper).
• Anthropic — scores transcript/scene data to suggest which moments make good clips.
• Stripe — billing.
• Postiz (self-hosted by us) — publishes your accepted clips to the platforms you’ve connected.
• OpenStreetMap (Nominatim) — turns a recording’s GPS metadata, when present, into a city/venue name for caption templates. Receives only the coordinate, never any content or account information.

Instagram and Facebook data specifically

If you connect an Instagram or Facebook account, we request only the permissions needed to publish content on your behalf to a Page or Instagram professional account you explicitly choose: listing the Pages/accounts you manage, and publishing a video you’ve approved. We do not request or access your personal messages, contacts, friends list, or any content beyond what’s needed to publish the clips you approve. We do not post anything to your connected accounts unless you’ve explicitly triggered that publish action in SceneStealer.

You can disconnect a connected account at any time from within SceneStealer, which revokes our access token immediately. You can also revoke access directly from your Meta account’s settings.

Data retention and deletion

We retain your uploaded recordings and generated clips for as long as your account is active, so you can continue to review and publish from them. See our Data Deletion Instructions for how to request deletion of your data.

Data security

Data is encrypted in transit (TLS) and at rest by our infrastructure providers. Access to production data is limited to the people who operate the service.

Children’s privacy

SceneStealer is a business tool and is not directed at, or knowingly used by, children under 13.

Changes to this policy

If we make material changes to this policy, we’ll update the date at the top of this page.

Contact us

Questions about this policy or your data: support@scenestealer.app.