Only when you Scoop.
Scoop analyzes video only when you explicitly invoke it. Watching a video, opening a page or leaving a tab open does not start analysis.
Privacy by design
Clear by default. Active only when you ask.
Your video stays yours.
Scoop looks only when you ask it to, processes what it needs for that search, and then stops.
You Scoop→selected area is processed→products are identified→frame data is discarded by default
The short version
Only what you choose.
Scoop focuses on the visual area needed for the product search. When a server or provider is required, the default is to send the minimum useful crop or structured product query, not your whole viewing session.
Temporary by default.
The default frame flow is capture, crop, process, discard. Scoop does not keep full videos or build a screenshot archive of what you watch.
Products, not people.
Scoop is built to identify products and visible objects. It is not designed to infer sensitive personal attributes from people in video.
The details
What Scoop processes, what it avoids, and what happens when you use the product.
What Scoop processes
When you ask Scoop to identify something, it may process the selected visual region, nearby permitted frames when they materially help identification, and product-search information derived from that selection.
What Scoop does not collect by default
Scoop is not designed to continuously inspect playback, build a browsing-history profile, keep full videos, or maintain a persistent archive of frames from what you watch.
Product learning and retained data
To make later searches more accurate and avoid repeating expensive analysis, Scoop stores derived product descriptions, visual attributes, identity hypotheses, matching evidence, confidence labels and product references. Records may also contain the selected video or content identifier, timestamp, pseudonymous session identifier, feedback, merchant-click information and performance metrics. These derived records can remain after the analysis session; they are not a continuous browsing log.
Third-party processing
When you request a search, selected frame images or visual crops and derived product descriptions may be sent through our Cloudflare service to configured AI processing providers such as OpenRouter, Google Gemini or other enabled vision services for recognition and verification. Product searches may also involve catalog, marketplace and search providers, including eBay, Etsy, SerpAPI or Brave, depending on availability. These services process the data needed to provide the requested result under their own applicable terms.
Access credentials and usage
The extension keeps a pseudonymous installation identifier and alpha access token in local browser storage. If an administrator signs in, it also stores an administrator session token. Scoop processes selections, feedback and outgoing shopping clicks to deliver the service, prevent abuse, troubleshoot errors and understand which results are useful. It does not use this information for creditworthiness or lending.
Retention and deletion requests
Frames are not intentionally retained as a general viewing archive by Scoop, but derived product, matching, telemetry and feedback records may persist in operational storage. AI providers and hosting services may have their own processing and retention practices. To ask about access or deletion of information associated with your use of Scoop, email contact@article6.org; we will review requests subject to technical, security and legal limitations. Specific retention schedules are being established for this alpha.
Operational data
Operational telemetry may include request IDs, timing, provider status and errors needed to keep the service reliable. Sensitive frame content is not logged by default.
Browser permissions
The extension should request only the permissions needed for the action you asked it to perform. When explicit capture permission is required, Scoop should explain why at the moment it is needed.
Protected content
Scoop does not bypass DRM, decrypt protected streams or defeat browser and platform access controls. If a permitted capture path is unavailable, Scoop stops.
Shopping actions
Scoop does not silently open merchant pages, manufacture clicks or turn a match into a purchase action without you choosing to continue.
Sponsored results
Paid placement stays clearly labeled and never gets to masquerade as an Exact match. Commercial relationships do not override match truth.
The waitlist is different
You submit it. We store it.
If you join the Scoop waitlist, the name, email, role and channel or company information you submit are stored so we can contact you about testing and launch access. That information is separate from video analysis.
Scoop is designed to notice the thing you chose, not everything around you.
This page describes current Scoop alpha processing and retention practices. We will update these disclosures as providers, retention periods, and product behavior change.
Questions: contact@article6.org
Updated November 12, 2026