Privacy policy

1. Data we process and why

Photos and tool inputs

When ChatGPT calls create_artwork, our service receives the inputs supplied for that call: the selected art style (FM line, scribble, circle stippling, or triangle art), drawing settings (such as seed, density, line width, size and colors), and whether to remove the background. If a photo is supplied, its reference contains a file identifier (file_id), a temporary download URL (download_url), and optionally its file name and media type (file_name and mime_type).

We use these inputs to validate the request, configure the studio and make the selected photo available to it. The service processes the references transiently; it does not fetch the photo, save the inputs in a database, or log request bodies. A count by drawing style is recorded separately for operational monitoring. The browser downloads the selected photo directly from OpenAI's file service and processes its image content locally, including optional background removal. Images may contain personal information, including faces or details visible in the scene. We do not perform identity recognition or create biometric identity profiles.

You can also choose a JPEG, PNG or WebP file from your device (up to 15 MB), or use ChatGPT's file picker when available. When ChatGPT provides its upload feature, choosing a local photo in the studio uploads the selected file to ChatGPT so it can be reused in follow-up requests. If that feature is unavailable, the chosen photo is processed locally without that upload. Only files you select are read.

The app does not require an Algorithmic Ink account, payment details or authentication. It does not request access to your entire chat history, other conversations, contacts, or files you have not selected. It receives the tool inputs ChatGPT supplies, rather than a copy of the entire conversation.

Tool outputs, studio updates and downloads

The tool returns a status (awaiting_photo or rendering_in_widget), style and label, drawing settings, background-removal choice, and the source file ID when present. These values and a short explanatory message are available to ChatGPT. The photo reference, including the temporary download URL, is returned in metadata intended for the studio rather than in the model-visible tool result.

As you use the studio, it sends ChatGPT rendering, completion, cancellation and settings-change status, current settings, the source file ID, and error text if a rendering error occurs. On completion it also sends a PNG preview with a longest edge of up to 512 pixels. These updates let ChatGPT understand the result and help with follow-up edits. Host-managed widget state can include those settings and statuses, the source file ID, file name and media type; the app does not put the temporary download URL in saved widget state.

When you press Download SVG or Download PNG, the studio passes the generated file contents, file name and media type to ChatGPT's download feature when supported. When the native download feature is unavailable, the embedded studio can upload the export to ChatGPT and offer an Open file link. A standalone studio can use a direct browser download. Full exports are not sent to Algorithmic Ink's servers. Downloaded copies remain on your device until you remove them.

Technical and support information

AWS hosts our integration website, tool service, application files and background-removal model. Handling network requests involves technical information such as the connecting IP address, user agent, request headers, requested path and time. For calls relayed by ChatGPT, some of this information describes OpenAI's connection rather than your device. We use service counts, timing, resource usage and operational logs to run, secure and troubleshoot the service.

Our application logs a count of validated tool calls by drawing style and a generic message for a failed request, not photo references, tool payloads or artwork. AWS runtime logs can also contain invocation IDs, timestamps, duration, memory use and runtime diagnostics. We have not enabled CloudFront access logging or S3 server access logging for this integration.

If you email us, we receive your email address, any name you provide, your message and attachments. We use this information to respond to support or privacy requests and keep necessary records of their resolution. Please do not send photos or temporary download links unless they are needed for your request.

We do not sell this data, use photos or artwork to train AI models, or use the ChatGPT studio's data for advertising, marketing profiles or cross-site tracking. The studio offers optional usage analytics described below. It does not load Google analytics scripts or advertising cookies. Browser caches may store application files and the background-removal model.

Optional studio usage analytics

Usage sharing is off by default. If you enable “Share usage analytics for this studio session”, the studio sends subsequent tool-result receipts, drawing completions and failures, export actions, drawing style, elapsed time, failure stage and application release to our backend, which forwards them to Google Analytics. Enabling sharing also records that the studio is open. Earlier actions are not replayed. An export event means an open or download request, not proof that a file was saved.

A random identifier and session timestamp exist only in browser memory for the opted-in session. We do not send photos, artwork, prompts, file identifiers, file URLs or error text to Google Analytics. We do not forward your IP address, browser headers or referrer; Google receives requests from our AWS backend. Advertising personalization and advertising user-data consent are set to denied. No analytics cookies or persistent analytics IDs are stored by this studio. These reports describe opted-in sessions, not unique people or installs.

Turn sharing off to stop future events. Closing or reopening the studio resets the choice. Turning it off does not erase events already sent. Google processes collected events according to the property's retention settings and its privacy policy. Aggregate reports may be retained longer than event-level data.

2. Who receives data

• OpenAI / ChatGPT: receives the tool outputs, uploaded photos when supported, studio updates, preview and download contents described above. OpenAI also hosts the selected ChatGPT files. Its handling of conversations, files and widget state is governed by your ChatGPT account or workspace settings and OpenAI's privacy policy.
• Amazon Web Services (AWS): provides CloudFront delivery, Lambda request processing, S3 static-file hosting and CloudWatch operational monitoring. AWS processes the tool request and response data and technical information needed to provide these services. Our backend runs in the United States (us-east-1); content delivery can involve AWS locations in other countries.
• Google Analytics: receives the optional usage events described above to help us understand reliability and feature usage. Processing may occur outside your country.
• The operator and email service providers: process support messages and privacy requests. The operator can access operational logs to maintain the service.
• Legal recipients: information we hold may be disclosed when required by law or necessary to address fraud, security incidents or protect legal rights.

We do not send photos to a separate image-processing provider. The background-removal model runs on your device and is served from our AWS-hosted assets. Using the service may involve processing outside your country, including by OpenAI and AWS.

3. How long data is kept

• Tool inputs and photo references on our server: processed in memory for the request and not deliberately retained after it. There is no app session database or stored history of tool inputs and outputs.
• Photos and artwork in the studio: held in browser memory while the studio is open, including to support regeneration. Closing the studio ends that active processing session; it does not delete copies already shared with ChatGPT.
• Operational logs: the integration's CloudWatch log group has a seven-day retention period. Aggregate service metrics follow AWS's metric retention schedule and do not contain photo files, photo references or artwork.
• Browser caches and downloads: application and model caches remain until the browser evicts them or you clear them. Downloaded artwork remains until you delete it, including any copies in your own backups.
• ChatGPT copies: OpenAI controls retention of uploaded files, conversations, previews, widget state and files handled by its download feature. The seven-day log period above does not apply to those copies. Their retention depends on OpenAI's policies and your account or workspace settings.
• Support correspondence: retained for as long as needed to resolve your request and keep necessary records of its resolution, security issues or legal obligations. There is no fixed automatic deletion period for support email. You can ask us to delete correspondence we no longer need.

4. Your choices and controls

• Leave usage sharing unchecked to use the full studio without Google Analytics. Disable it at any time to stop future events.
• Choose which photo to provide, or do not provide one. Avoid including sensitive information you do not want processed or shared with ChatGPT.
• Use Cancel drawing to stop an active rendering job, or close the studio to end its active session. Cancellation does not undo an upload or updates already sent.
• Turn off Remove background to skip local background-removal processing and its model download. This does not disable the studio's sharing with ChatGPT.
• Disconnect or disable Algorithmic Ink in ChatGPT to stop future use. Disconnecting does not delete existing conversations, files or exports.
• Use ChatGPT's conversation, file and data controls, or your workspace administrator, to manage copies held by OpenAI and applicable model-training settings. Algorithmic Ink cannot delete data from your OpenAI account.
• Clear browser site data to remove cached assets and delete downloaded files from your device if you no longer want them.
• Email algorithmicink@mahfouz.ca to ask about, access, correct or delete personal information we hold, or to raise a privacy concern. Depending on where you live, you may also have rights to object to or restrict processing, request portability, withdraw consent or complain to your privacy regulator. We will assess requests under applicable law and may need information to verify that a request relates to you. We cannot retrieve or delete a server-side photo library because we do not maintain one.

The ChatGPT studio currently has no separate switch to disable completion previews or status sharing with ChatGPT. If you prefer photo processing without the ChatGPT integration, you can use the standalone studio at algorithmicink.ca. That website has its own privacy notice and uses website analytics; its analytics are not loaded in this ChatGPT studio.

5. Security and policy updates

We use HTTPS, restrict access to hosting resources, and avoid retaining tool payloads in our application logs. No transmission or storage method is completely secure.

We will update this policy when the app's data handling changes and revise the date above. Contact the operator at algorithmicink@mahfouz.ca if you have questions about a change.