Privacy Policy

AskAVideo is a ChatGPT plugin that helps ChatGPT retrieve available YouTube transcripts through your Chrome session so it can discuss a video with you. The public plugin includes an account and video-credit system. It does not operate a hosted transcript service or create an advertising profile. The publisher operates only the account, billing, measurement, and private-research systems described below.

Information processed

When the plugin is used, ChatGPT processes the YouTube URL and your question and uses its Chrome capability to access the supplied video page. The workflow may read the title, channel, duration, available caption language, and transcript text with timestamps from that page. The plugin does not ask for or receive your YouTube password.

How information is used

The URL, metadata, and transcript are used within your ChatGPT conversation to answer the request you made and create timestamp links. The plugin does not send the URL, metadata, question, or transcript to the publisher's measurement or private-research databases.

Plugin account and video credits

To provide the monthly allowance and purchased credits, the account service stores your email address, an internal account identifier, OAuth authorization records, plan and credit balances, and the minimum billing-provider identifiers needed to keep access current. Sign-in uses a short one-time code sent to the email address you provide. Authentication tokens are stored only as cryptographic hashes.

Before transcript retrieval, ChatGPT sends only the video's 11-character YouTube ID to the account service. The service converts it immediately into an irreversible keyed fingerprint and stores the fingerprint, UTC month, and charge result. It does not receive or store the video URL, title, transcript, excerpts, your question, ChatGPT's answer, or your conversation history.

Payments

Polar is the merchant of record and processes checkout, payment, tax, receipts, refunds, and subscription events. AskAVideo stores only the minimum customer, subscription, transaction, plan, and credit records needed to grant or revoke access. Payment card details are handled by Polar and are never received by YouTube Conversation.

When checkout is created through our Cloudflare Worker, the connecting IP address is forwarded to Polar so it can determine the correct country, currency, and applicable taxes. AskAVideo does not store that IP address in its billing database.

Service providers

The workflow relies on OpenAI/ChatGPT and its Chrome capability, YouTube to supply video pages and captions, Cloudflare to host the account service and public pages, an email provider to deliver one-time sign-in codes, and Polar to process billing. Their handling of data is governed by their own policies.

Anonymous website measurement

The public site uses Cloudflare Web Analytics to count page visits. Our first-party measurement also records the first landing path, public campaign tags, the referring site's hostname when the browser supplies one, and whether an Install link was selected. It never records the full referring URL or its query, and it honors Global Privacy Control and Do Not Track. We do not store your IP address, YouTube URL, ChatGPT prompt, transcript, answer, or account identity. This measures aggregate acquisition and install interest; it does not tell us whether you completed the installation.

Private activation research

Invited adults may volunteer for a private uncoached activation study. The research page does not load ordinary website analytics and never asks for or stores a name, email address, YouTube URL, question, transcript, answer, or ChatGPT account identifier as a database field. The browser-tab-only recording may visibly contain the public video URL, test question, transcript excerpts, and product answer. Participants must hide account/sidebar details and keep camera and microphone off. The form accepts only an encrypted locator to a recording restricted in Google Drive.

Recruitment coordination is separate from the pseudonymous outcome data. It temporarily stores the candidate's name, public professional email address and source URL, segment, exact initial and follow-up copy, a fixed screener template containing only a placeholder, delivery state, timestamps, and Gmail message and thread identifiers. Gmail is the email provider. Only email in the verified recruitment thread is inspected transiently to classify a generic reply, an explicit positive reply of test, or an unsubscribe. The system stores that classification, not a full reply body, snippet, AI-generated summary, or raw private-invitation token. The fragment URL is verified against its stored hash and substituted only in the Gmail request.

The research system stores one-way hashes of random invitation and recruitment-reference values, structured consent and outcome codes, one temporary recruitment-candidate UUID, and an AES-GCM-encrypted locator for that restricted recording. The participant restricts the Drive file to their own Google account and [email protected]. Google processes the recording as the storage provider under the participant's Drive account controls; the research administrator can retrieve the encrypted locator. The sharing checkbox is a participant declaration, not automatic proof. Before viewing content, the administrator manually inspects Drive permission metadata without downloading the recording and records only pending, restricted verified, or unsafe rejected. Unsafe rejection deletes the locator immediately, and coding is blocked until restricted access is verified. The Worker has no Drive OAuth and cannot perform this check automatically. We delete the locator by day 14. The participant controls the Drive source and must revoke [email protected] access and delete the original by day 14; coding never extends that deadline. A participant's identity, recording, or quote will not be used publicly without a separate explicit opt-in, and this study has no public-quote submission path. Candidate-linked contact and source details, message copy, delivery metadata, and Gmail records are purged 30 days after the cohort closes, regardless of Gmail availability. If permanent Gmail deletion has not yet succeeded, deletion-only provider message identifiers may remain solely until Gmail deletion succeeds; they contain no email, name, segment, subject, body, or candidate link. An opt-out suppression address is retained only to prevent future contact. Withdrawal deletes the locator, screening outcome, pseudonymous research session, outcome coding, return checks, and recruitment bridge immediately and cancels pending research follow-ups. Cloudflare necessarily processes network requests to operate the form, but the research application does not persist or intentionally log IP addresses or request metadata.

Your choices

Do not use the plugin for a URL or browser session you do not want ChatGPT to process. You can review or disable browser access in ChatGPT settings. The publisher does not maintain a transcript database. To request deletion of your plugin account or exercise a privacy right, contact [email protected]. Billing records that Polar or applicable law requires may remain for the required period.

Changes and contact

Material changes will be reflected on this page with a new update date. For privacy questions, use the contact route on the Support page.