Privacy Policy

Last updated: September 9, 2026

Jarvis Content Operations ("Jarvis," "we," "the application") is a single-user, locally-run capability inside the Jarvis personal AI assistant. This policy explains what data it accesses and stores when connected to TikTok, YouTube, or Meta (Instagram/Facebook) on its owner's behalf.

Who this applies to

Jarvis Content Operations is operated by and for one individual user, on their own computer. It is not a public multi-user service, does not create accounts for other people, and does not knowingly collect data from anyone other than its own operator and the public content it researches.

What we access

What we store

• OAuth tokens for each connected platform, stored locally on the user's own machine — never transmitted to any third party beyond the platform that issued them.
• Content records — the scripts, research sources, render metadata, and publication history Jarvis itself produces — stored locally, on the user's own machine.
• Analytics Jarvis reads back from each platform about the user's own published content (views, engagement, etc.), stored locally for the user's own review.

None of this data is uploaded to a third-party server, sold, shared with advertisers, or used to train a public model.

What we don't do

• We do not access any other user's private data on any platform.
• We do not read, store, or act on another person's direct messages, contacts, or private posts.
• We do not use platform data for advertising or resell it in any form.

Security of Google User Data

This section describes, specifically, how Google user data (YouTube channel information, video metadata, and analytics) is protected. Every statement below reflects the application's actual implementation, not a general assurance.

• Encrypted in transit. Every call to a Google API — uploading a video, reading channel or video data, or querying analytics — is made exclusively over HTTPS/TLS. The application never sends Google user data, or the credentials used to access it, over an unencrypted connection.
• Authorization tokens are never placed in a URL. The OAuth access token is sent only in a standard Authorization: Bearer HTTP header on each request — never as a URL query parameter, where it could otherwise be captured in a proxy or web-server access log.
• Local storage only, scoped to one device. The OAuth access and refresh tokens Google issues, and the OAuth client credentials for this application, are stored in a local file inside the application's own folder on the single computer where the user installed and authorized it. They are never transmitted to any server operated by us or by any third party.
• Operating-system-level access control. That local file inherits the normal file permissions of the user's own operating-system account — on a typical single-user computer, only that account can read it. The application does not currently apply an additional layer of encryption to the file at rest, and does not claim to.
• Never written to logs or shown in the interface. The application's own logs, error messages, and on-screen status displays are built to describe a token's presence (connected/not connected, when it was obtained, which permissions it carries) without ever including its actual value. A person operating the application can see that YouTube is connected and see real data it has read back, but never the underlying token.
• Limited to the scopes actually granted, for the purpose stated at authorization. Google user data is used only to (a) upload videos the user has explicitly approved, to the user's own connected channel; (b) read back that channel's own identity and the status/metadata of videos the user's own account uploaded; and (c) read analytics for the user's own published videos. It is not used for any other purpose.
• Never sold, shared, or used for advertising. Google user data is never sold, never shared with any third party, never used to serve or target advertising, and never used to train a general-purpose or public machine-learning model.
• Human review, not automated profiling. Any reading of Google user data described here is either directly requested by the user or performed by automation acting on the user's own explicit, standing instructions — never automated decision-making about the user based on that data for any purpose beyond running this application's own stated features.
• Revocation. The user can revoke this application's access at any time from their Google Account's own third-party access settings. Doing so takes effect immediately on Google's side; the next time the application tries to use the connection, it fails and honestly reports that re-authorization is required — it never continues to access data after a revocation succeeds. The application also provides its own "Disconnect" control, which immediately deletes its local copy of the token without requiring the user to leave the app.

Artificial Intelligence & the Google API Services User Data Policy

This section states, specifically, whether any data obtained from Google APIs is ever provided to an AI or machine-learning system, and if so, how.

• Google/YouTube user data is never sent to Gemini, or to any other AI or machine-learning system, for any purpose. Concretely: none of the following is ever included in a request to Gemini or any other AI provider — a YouTube channel's identity or statistics, a video's metadata or upload/readback status, YouTube Analytics data, or any value derived from them. This is not merely a policy statement — it reflects how the application is actually built: every AI call in Jarvis's Content Operations pipeline runs strictly before a video is ever uploaded (researching a topic, drafting a script, fact-checking a claim, planning a storyboard, or optionally generating an illustrative image), and operates only on content Jarvis itself is authoring — a research query, a script draft, a brand's own content guidelines. None of these calls accepts or has access to YouTube channel, video, or analytics data. The one system that processes real YouTube analytics after a video is published — Jarvis's own internal comparison of how different videos performed — is plain, deterministic arithmetic (comparing view counts and computing percentage differences in code); it never calls Gemini or any other AI system.
• Complete list of AI/ML integrations used at runtime: Google Gemini, called directly through Google's own generative-AI SDK using Jarvis's own API key, for purposes entirely internal to content creation, before publication — researching and drafting topic ideas, drafting and fact-checking video scripts, planning a video's visual storyboard, and optionally generating illustrative (non-photographic, non-likeness) images when no suitable licensed stock photo can be found. None of these calls ever receives Google/YouTube user data, as described above. No other AI, machine-learning, or generative service is used to process any data obtained through the YouTube Data API or YouTube Analytics API.
• No multi-model service, model aggregator, or gateway is used. Jarvis calls Google's Gemini API directly; no request is routed through a third-party model router, proxy, or multi-provider aggregation service, and no upstream or downstream model besides Gemini is ever involved.
• No self-hosted or offline AI model is used. Content Operations does not run any local or offline machine-learning model — every AI call described above is a live call to Google's own hosted Gemini API.
• Limited Use compliance. The use of raw or derived user data received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements. In particular: Google user data obtained through the YouTube Data API and YouTube Analytics API is used only for the user-facing features described in this policy (uploading the user's own approved videos, and reading back their own channel/video status and performance); it is never transferred to any AI or machine-learning system; it is never used to train, improve, or fine-tune any model — including Gemini — of any kind; and it is never sold, shared with advertisers, or used for any purpose beyond providing this application's own stated features to its single owner-operator.

Data retention and deletion

All data described above lives locally on the user's own machine for as long as they choose to keep it. Disconnecting a platform (via the in-app "Disconnect" control, revoking the app's access in that platform's own account settings, or removing the credential file inside Jarvis) stops all further API access immediately; the user can delete any locally stored record, including the token file itself, at any time.

Changes to this policy

If this policy changes, the "Last updated" date above will change accordingly.

Contact

Questions about this policy or a data request: support@jarviscontentops.com