Privacy policy.

Information Blip handles

• Provider settings and authentication: your chosen connection type, provider, model, localhost endpoint, and feature preferences are stored in Chrome's local extension storage. Cloud API keys use local storage. Optional Local server tokens use session storage unless you explicitly choose Remember on this device. API keys authenticate requests to the corresponding provider. They are not supplied to page scripts or included as content in AI prompts.
• Page and video content: Blip reads relevant URLs, titles, readable page text, captions, and video metadata. When captions are unavailable, it may use relevant video comments, a visible-tab image when supported, and provider search features to construct an answer. These fallbacks may produce less reliable results than a transcript.
• Documents and spreadsheet cells: when used with Blip, PDF text, CSV/TSV/XLSX cells and authorized Google Sheets ranges may be read to provide summaries and answers. These inputs, screenshots and questions may contain names, contact details, personal communications, health, financial or location information. Relevant content can be sent to your selected AI provider; Blip does not create a separate advertising or sensitive-interest profile from it.
• Conversations: your questions, relevant conversation context, extracted source text, generated summaries, chapters, and answers are processed for the features you use. Local chat history and content caches allow conversations to resume.
• Images: visible-tab screenshots used for image analysis, and supported captionless-video fallback, are sent to your selected provider when an image-capable model is used. Screenshots may contain personal information visible on the page.
• Diagnostics: Blip keeps a bounded local log of operational events and errors and attempts to redact common secrets. You can clear it or export it in Settings. Exporting saves a file locally; sending that file to someone is a separate action you control. Redaction is not a guarantee that every piece of sensitive page or error information is removed.

Blip does not read your full Chrome history database, record a general keystroke or mouse-activity stream, or use your data for advertising. A page's readable content, your questions, or a screenshot may contain personal or sensitive information; using Blip on that content can send it to your selected AI provider.

AI providers and other network services

In API mode, requests go to the service you choose: Google Gemini AI Studio, OpenAI, OpenRouter, or NVIDIA NIM. In Local mode, requests go to the configured localhost Ollama or LM Studio API, including an explicitly configured local tunnel port. The request can include page/video content and metadata, your question, relevant chat history, and image inputs where supported. Provider search features may process search queries and relevant request context. Model-list and capability requests also contact provider endpoints. OpenRouter may route requests to downstream model providers.

Each provider's privacy policy, account settings, retention rules, model availability, and charges apply to its processing. Blip cannot delete copies already retained by those services or promise that a provider never uses data for model improvement. Review the policies and settings for the service you use.

• Google Gemini AI Studio: https://policies.google.com/privacy
• OpenAI: https://openai.com/policies/privacy-policy
• OpenRouter: https://openrouter.ai/privacy
• NVIDIA NIM: https://www.nvidia.com/en-us/about-nvidia/privacy-policy/

YouTube is contacted to retrieve video information, captions, and relevant comments. Other sites may receive requests for caption resources needed to understand their current page. Network services receive ordinary connection information such as your IP address and request metadata.

Optional Google Sheets connection

You choose Connect Google Sheets to authorize Blip through Google OAuth. Blip requests the https://www.googleapis.com/auth/spreadsheets.readonly scope, which permits seeing all Google Sheets spreadsheets your account can access. This scope is broader than a single worksheet or cell range. Blip uses it to read selected spreadsheets opened in, or linked from, a page where you use Blip. It does not crawl Google Drive or enumerate all your spreadsheets. The connection does not permit Blip to edit, create or delete spreadsheets. It does not request access to Gmail, Calendar or your Google password.

Blip retrieves spreadsheet titles, worksheet names, grid metadata and bounded cell ranges. Cells can include text, underlying typed values, formatted display text, formulas and Google-calculated results. These values may contain personal or sensitive information. Worksheet and range references identify the evidence; unread ranges remain explicitly limited.

Relevant spreadsheet content, your question and conversation context are sent directly to your selected AI provider when Blip generates a summary or answer. The provider receives this information to generate the response. Blip does not send Google OAuth tokens, Google passwords or Google authentication cookies to AI providers. Choose the account, sheets and AI provider you are comfortable using together. Provider terms, retention and account settings apply as described above; Blip cannot erase data already retained by a provider.

Google OAuth tokens are managed by Chrome's identity token cache and used in the extension's background worker for requests to Google. Blip does not write raw tokens into its settings, conversations, source indexes or diagnostic logs, and does not provide them to webpage scripts. A local Boolean connection flag is stored to remember whether you connected. Checking connection status does not read spreadsheet cells. Blip does not request a separate Google profile, email-address or contacts scope for this connection.

Normalized spreadsheet evidence is held in bounded worker memory with a five-minute expiry while the worker runs. It is cleared when the owning page navigates or closes, when the connection is disconnected, or when the worker stops. Blip does not create a separate persistent collection of your Google workbooks. Questions, relevant excerpts and generated answers retained as conversation history follow the local storage rules below. Google receives authorized API requests and normal connection metadata, such as an IP address and request details.

Use Google Sheets connection → Disconnect Google Sheets to stop subsequent reads, invalidate worker evidence and clear Chrome's cached authorization. Disconnecting does not delete existing chats, exports or data already sent to a provider. You can separately remove Blip's app grant through your Google Account connections, and remove conversation history through Blip's chat controls, Sources → Forget this page, or the extension's stored data. Disconnecting is separate from revoking the app grant in Google Account connections. Google testing restrictions, sheet-sharing permissions and organization policies can limit access; Blip does not bypass them. Other Blip features do not require this connection.

Other spreadsheet and online document sources

When selected by the page's extraction flow, accessible CSV, TSV, XLSX and PDF resources may be downloaded for local parsing by code bundled with the extension. Readable HTML tables may also be inspected. Additional resource hosts require site access where necessary. These document downloads omit cookies and credentials; login-protected, encrypted or unsupported files can remain unreadable. Blip does not execute spreadsheet formulas. Relevant extracted values and document excerpts can be sent to your selected AI provider through the normal summary and answer flow. Downloads and parsing have size, time and coverage limits.

Optional local transcription

Automatic YouTube transcription is disabled by default. Local transcription requires separate helper setup and a compatible computer. The extension creates a platform-specific installer ZIP from packaged files. You extract and run its unsigned script yourself. Setup downloads external runtimes and dependencies, including Python tooling, Node.js, speech libraries, and the Windows speech runtime where applicable. Model downloads contact Hugging Face and its download infrastructure; setup also uses upstream software distribution services such as GitHub, Node.js, and Python package services. These services receive normal download-request information. Blip does not send them your video transcript for transcription.

Reconnect helper can generate a small transcription registration-repair ZIP. You extract and run it yourself. It reuses an existing runtime and model files without downloading them, and registers the current extension ID while keeping previously registered valid IDs for the same helper launcher.

The helper downloads video audio from YouTube and performs speech recognition on your computer. Temporary audio is removed after processing, errors, or normal cancellation. An abrupt operating-system or process failure can leave temporary files requiring local cleanup. Downloaded models and runtime files remain on disk for reuse.

Local speech recognition does not make AI chat offline: the resulting transcript is sent through the normal selected-provider flow for summaries and answers.

Use my YouTube session is selected by default within the optional local transcription settings. Cookie access still requires Chrome's separate permission grant. When that option is selected, permission has been granted, and local transcription is used, Blip reads YouTube-domain cookies from the current tab's Chrome cookie store. It passes them through the local native messaging connection to an in-memory, domain-scoped cookie jar for authenticated YouTube audio requests. Cookies are not exported, saved by the helper, included in AI prompts, or read directly from Chrome's on-disk cookie database. Model downloads do not use these cookies. Disabling the session option stops their use for subsequent transcription jobs; Chrome's extension settings allow you to manage granted permissions.

Optional non-YouTube video source access

When you invoke Blip on a webpage, it can inspect permitted embedded frames and caption tracks associated with the page. If another host is needed, Review site access lists the discovered hosts in an extension-owned page. You grant or revoke each host separately. The extension declares optional HTTPS host support to accommodate caption CDNs discovered at runtime, but does not request access to all websites. Host grants remain until revoked here or in Chrome's extension settings.

Optional frame-discovery (webNavigation) permission may also be requested when Chrome cannot enumerate the page's frames. Chrome may describe this as browsing history access. Blip uses it only to identify frames in the tab where you invoked the feature; it does not subscribe to navigation events, collect browser-wide history or crawl other tabs.

Caption requests omit cookies and authentication headers. Site-issued signed caption URLs may be used transiently to retrieve the caption content. Resource source-URL metadata removes credentials, query parameters and fragments. Source indexes and pending access requests are bounded and held in memory; pending review tickets expire or become invalid after page/document changes. Relevant caption excerpts can be sent to your selected AI provider to answer your question. Caption acquisition does not record tab audio or bypass protected player access.

Background non-YouTube audio transcription

When local transcription is enabled, activating Blip on a non-YouTube page can prepare a supported video's audio without playback. Usable captions are checked first. A single captionless video prepares automatically; pages with multiple videos show checkboxes in Sources before downloading. That same panel supports additional selections and cancellation. A small status reports background work. Questions can use finished sections while remaining audio is still processing; partial coverage is explicitly identified.

The local helper downloads accessible media directly from the source website or its associated public media hosts, decodes local files and performs recognition using installed models. It does not record tab playback or your microphone. Temporary media is removed after processing, failure or normal cancellation; operating-system failure can leave files needing local cleanup. Switching tabs keeps preparation running; navigating or closing the original tab cancels its unfinished jobs. Supported formats/providers, duration, byte, queue and processing limits apply. Protected or unsupported media remains unavailable.

Downloads are attempted without cookies first. If authentication is needed, the existing trusted site-access page can request Chrome's cookie/website permission and your additional approval for the specific site. This host grant covers that registrable site and its subdomains to expose parent-domain login cookies; unrelated hosted tenants remain separate. The helper receives applicable cookies from the original tab's Chrome cookie store in a job-local memory jar. Domain, host-only, path, HTTPS and applicable partition boundaries are maintained. Cookie values are not saved to cookie files, included in prompts/logs or sent to an AI provider, and are sent only to approved associated request origins, not every subdomain covered by Chrome's permission. You can stop session reuse through the same site-access flow or revoke Chrome permissions. Cookie/access changes invalidate affected private transcript cache entries. This is access reuse for supported downloaders, not a guarantee that every signed-in player is downloadable.

Completed non-YouTube transcripts are cached in trusted extension storage on this device for seven days, subject to additional size/count eviction. Expired entries cannot be reused; an alarm cleans them while Chrome runs and cleanup also runs at the next worker startup/access. Source/metadata validation is required for reuse. In-progress transcript sections stay in memory and are discarded on cancellation/failure. Private-window transcripts do not enter the ordinary profile cache. Clear non-YouTube transcripts in existing local-transcription details clears saved and derived evidence. It does not delete conversation history or provider records. Transcript excerpts reach your selected AI provider through the normal answer/summary flow; downloaded audio does not go to that provider.

PDF reader and annotations

You explicitly choose or drop a local PDF into Blip reader. The selected file is rendered and its text is indexed locally using bundled PDF.js code. Raw PDF bytes stay in reader/worker memory and are not saved in extension storage or uploaded as a PDF file. Reloading the reader requires choosing the file again. Relevant extracted text, document title, questions and summaries use the normal selected AI provider flow when you request a summary or answer. Saved chats follow the existing local conversation policy and are matched using a content digest, not a filesystem path. The original tab address is temporarily retained in extension session storage so Back to original PDF can return to the viewer; it is not included in prompts or diagnostic logs by the reader. This workflow does not require access to other local files or a file-URL permission grant.

For larger documents, extracted page text and layout metadata that exceed the in-memory window may be cached locally in extension IndexedDB for the current reader session. Closing the reader triggers removal of its cached pages. Cached text expires after 24 hours; reads do not extend expiry, and expired records cannot be reused. Cleanup runs at worker startup and hourly while Chrome runs. If the browser is closed or cleanup fails, physical deletion occurs at the next successful cleanup opportunity. Settings → Clear PDF cache removes cached PDF text and cancels active PDF work; reconnect open files to continue. It keeps original files and saved chats. Raw PDF files are not stored in this cache.

Supported HTTPS PDF links can also be opened in the reader after you approve access to the document's origin. These downloads omit cookies and credentials and remain subject to size and access limits. The document bytes and unsaved highlights, underlines, notes or annotation removals are kept in reader memory. Save As or Download PDF writes an annotated copy to the location you choose; Blip does not automatically upload that PDF or its annotations to a storage service. Saved PDFs and exported chats remain on your computer until you remove them. Closing or reloading can discard unsaved annotation changes.

Read aloud and optional AI voices

Automatic voice passes the visible completed reply or fixed Settings sample to an explicitly selected eligible local system voice through Chrome text-to-speech. Remote and third-party extension voices are excluded. Reading an existing reply requires no provider key. Playback starts on request; automatic reading is off. Voice/model/language/rate preferences are stored locally. A temporary Boolean session marker supports interrupted-playback cleanup without retaining text or audio.

Browser Kokoro downloads verified model data only after an explicit action and optional Hugging Face host grant. Executable JavaScript, WASM and phonemizer assets are bundled. Verified models and voice data remain in extension Cache Storage for reuse until removed or evicted. Normal q8 inference uses CPU; the separate diagnostic page can explicitly compare CPU/GPU using a fixed sample. Text and generated audio are not sent to a TTS service or logged. Generated AI audio may be retained locally as described below.

Native Kokoro/Pocket speech uses the separately installed com.blip.local_tts helper on compatible computers. Setup creates an unsigned installer ZIP that you extract and run. It downloads external Python tooling, a native engine, phonemizer and dependencies; Windows may need the official Microsoft C++ runtime installer and normal administrator approval. Model/voice/reference downloads use Hugging Face infrastructure. Upstream distribution services receive ordinary download metadata. Inference uses verified local files and sends no reading text or audio to an external speech service. Models and runtime files persist locally.

You can select and download multiple Kokoro voices across supported languages. The helper reuses shared model files and retains successfully installed voices if you cancel or interrupt a later download. Removing a model in setup also removes preset voice files no longer needed by other installed models; custom recordings have a separate delete control. Read-aloud text and generated audio pass between the extension and the local helper through Chrome native messaging for playback on your computer. Once the required files are installed, speech synthesis runs locally without a cloud speech fallback.

For a custom Pocket English voice, you explicitly select a permitted 1–30 second recording. Chrome decodes/resamples it locally and the helper stores a normalized 24 kHz mono PCM16 WAV copy plus voice metadata in its private data directory. The original selected file is untouched. Derived reference conditioning stays in worker memory; persistent upstream latent caching is disabled. Delete selected custom voice removes the stored normalized copy and metadata. Model removal does not remove separately managed custom recordings. Recordings are not uploaded.

Prepared browser and native AI read-aloud audio is saved in extension IndexedDB on this device for up to seven days after its last playback; starting playback renews that period. Opening a page alone does not renew it. The cache is limited to 256 MiB of audio and 512 reply/voice variants, with the least recently played entries removed first. Only completed, validated audio chunks are retained. Returned pages and restored conversations can reuse them without sending text to a speech service or repeating synthesis. Settings test samples, automatic system voices, and private-window playback are not saved in this cache. Audio is matched to its saved reply, spoken text, model, voice, language, speed, and synthesis version. Expired or removed audio is generated again only when requested. Clearing a chat, deleting its chat tab, forgetting its page, and saved-chat expiry remove its associated audio. Clear saved audio in Read aloud settings clears this cache without removing conversations or installed models. Expired entries cannot be reused; physical cleanup runs on access, at worker startup, and hourly while Chrome runs. Audio stays separate from chat exports and is removed with Chrome-managed extension data.

Microphone assist

Dictation requires an explicit Moonshine English model download and a microphone grant in a visible extension-origin setup tab. The grant check closes its temporary stream immediately. Recording starts only when you click the chat microphone. Audio is processed locally in an extension offscreen context, is not uploaded, saved or logged, and does not pass through the AI provider. Provisional text stays outside chat history. Stop ends capture before final inference; Cancel and owner teardown discard speech previews. Sessions are limited to 60 seconds.

Final text is inserted into the editable unsent draft. Only pressing Send submits it through the normal selected-provider flow; sent questions then follow normal chat-history retention. The microphone model stays in a separate extension Cache Storage inventory until removed in setup or evicted. No native helper is required.

Storage, retention, and your controls

Ordinary page source indexes are kept in worker memory, with a five-minute physical expiry while the worker runs and cleanup when their owning tab navigates or closes. They disappear when the worker stops. Saved chat text is separate: up to 30 sessions are retained until deleted or evicted by the session limit by default. Settings can opt into expiry 30 or 90 days after a session's last update. Expiry is enforced on access, worker startup and hourly while Chrome runs. New saves have 4 MiB per-session and 16 MiB total limits; a rejected save shows a notice and preserves previously saved data. Older stored chats are not silently shortened to meet these new byte limits.

Saved chats and JSON exports include a compact historical source inventory, capture time and revision references, not a raw page archive. Restored replies are historical; Refresh sources reads the current page without rewriting earlier answers. Sources → Forget this page removes that session's local chats and associated source/derived caches. Exported files and copies already sent to a provider remain separate. Locked or inactive HTML table views do not supply numeric answer evidence. Diagnostic URLs have credentials, query strings and fragments removed before new log entries are saved.

Extension settings, API keys, chat history, bounded content caches, model capability information, widget preferences, setup state, and diagnostic logs are stored locally using Chrome extension storage, not Chrome Sync. Prepared AI read-aloud audio additionally uses local extension IndexedDB; browser speech/microphone model data uses local extension Cache Storage; native helper models and voice references use separate OS data directories. Blip does not add its own encryption layer to this local storage; device and browser access controls protect it. Cloud API traffic uses HTTPS. Local model APIs use HTTP or HTTPS only on localhost. For remote models, your external SSH client or forwarding tool carries traffic from that local port to your authorized server; Blip does not establish or independently verify the tunnel. Native messaging connects processes on the same computer.

You can change or remove provider keys in Settings, clear diagnostic logs, and clear the learned context-window cache. Clearing that context-window cache does not delete all transcripts or conversations. Removing the extension removes its Chrome-managed extension storage. Exported chat or debug files remain wherever you saved them.

The native helper is installed separately and remains after the extension is removed. Its runtime is at ~/.local/share/blip-local-stt on macOS or %LOCALAPPDATA%\Blip\local-stt on Windows. To remove it, remove that runtime directory and its native messaging registration: ~/Library/Application Support/Google/Chrome/NativeMessagingHosts/com.blip.local_stt.json on macOS, or HKCU\Software\Google\Chrome\NativeMessagingHosts\com.blip.local_stt on Windows. Downloaded speech models remain separately in your Hugging Face cache until you remove the relevant model files. Shared caches or tooling used by other applications should be managed separately.

The separate speech helper is installed at ~/.local/share/blip-local-tts on Mac or %LOCALAPPDATA%\Blip\local-tts on Windows. Its model/normalized reference data is at ~/.cache/blip-local-tts or %LOCALAPPDATA%\Blip\local-tts-data. Close playback, remove the helper directory and its com.blip.local_tts registration from ~/Library/Application Support/Google/Chrome/NativeMessagingHosts on Mac or HKCU\Software\Google\Chrome\NativeMessagingHosts on Windows to uninstall. Remove the speech data directory separately if desired. Repair retains model data. Removing the Chrome extension does not uninstall either native helper or its data. Browser microphone and TTS models can be removed separately from their setup pages.

Deleting data and stopping access

• Saved conversations and source evidence: use the chat controls or Sources → Forget this page. Forgetting a page removes that session’s stored chats and associated source/derived caches; it does not erase other sessions or exported files.
• Google Sheets access: disconnect in Blip’s Google Sheets connection settings to stop future reads. Separately revoke Blip in Google Account connections to remove the Google authorization grant. Delete existing conversations separately.
• Provider keys and local diagnostics: remove API keys and clear diagnostic logs in Settings. Local server setup includes Clear server token and connection results, which removes the saved server token. Disconnect an external tunnel using its own client; revoke localhost host access in Chrome extension settings if desired. Uninstalling the extension removes its Chrome-managed extension storage; exported files and native helper data require separate removal.
• Non-YouTube transcripts, speech models and custom voices: use the relevant clear, remove or delete controls described in this policy. Native helper runtimes and data persist after the extension is uninstalled.
• Data sent to AI providers: contact the provider or use its account controls for deletion requests. Clearing Blip’s local storage cannot remove a provider’s retained copies.

The developer does not operate a server-side collection of your Google Sheets cells, chats or API keys. For privacy requests or help with deletion controls, contact exergyoflife@gmail.com. Include only the information needed for support; do not send API keys, Google tokens or sensitive documents.

Permissions and activation

• activeTab: grants temporary access to the current tab after a user gesture for toolbar activation and visible-tab screenshots. On ordinary HTTPS pages, click the Blip toolbar icon to activate the widget. Browser-protected pages are unsupported.
• storage: saves settings, provider keys, history, caches, local setup state, and diagnostic logs on your device.
• unlimitedStorage: supports long transcripts, summaries, conversations and persistent verified browser speech/microphone model caches. Caches/logs have application-level bounds; native helper models are stored separately.
• tts: reads completed replies with an eligible local system voice.
• identity (optional): obtains Google OAuth tokens after you choose Connect Google Sheets. Only read-only spreadsheet access is requested; Google passwords are not supplied to Blip.
• Google Sheets API host (optional): permits the background worker to contact sheets.googleapis.com for authorized spreadsheet reads after connection.
• webNavigation (optional): identifies embedded frames in the specific tab where you invoke Blip when ordinary frame discovery is insufficient. It is not used to monitor navigation across websites.
• Additional resource hosts (optional): retrieve selected caption tracks, embedded sources and supported online documents from specific hosts you approve through site access.
• offscreen: supports local inference workers, speech playback and authorized microphone capture in an extension-owned document outside the web page.
• alarms: schedules cleanup of expired locally cached non-YouTube transcripts and checks optional saved-chat expiry while Chrome runs.
• scripting: injects the packaged interface and content extractors into authorized tabs, seeks YouTube timestamps, and hides/restores the widget for screenshots.
• YouTube host access: enables the widget and caption, metadata, and relevant comment retrieval on YouTube while Blip is enabled. Caption retrieval may use the page's existing YouTube session. Optional native-helper cookie access is described above.
• AI provider host access: permits direct requests to the Google Gemini, OpenAI, OpenRouter, and NVIDIA endpoints when those services are used.
• Local model host access (optional): permits discovery, connection tests and inference at the localhost endpoint you configure for Ollama or LM Studio. HTTP access to 127.0.0.1 and localhost is requested only when needed. Local requests omit cookies, reject redirects and validate the configured port and API path; Local mode does not grant access to public or LAN model-server addresses.
• nativeMessaging (optional): connects to the separate local transcription and read-aloud helpers for runtime checks, model/reference management, speech recognition and synthesis.
• downloads (optional): saves transcription/speech helper installer or transcription reconnect ZIPs generated from bundled files and reveals them in your download folder. You must extract and run them separately.
• Hugging Face hosts (optional): allow explicitly requested browser model-data downloads from huggingface.co and *.hf.co.
• Microphone browser grant: authorizes local capture after visible setup and an explicit chat microphone click; page scripts receive no microphone grant.
• cookies (optional): enables the YouTube session option and separately approved per-site session reuse for supported non-YouTube downloads.

Blip does not request permanent host access to all websites. Its automatically registered content scripts run on YouTube; ordinary pages use temporary active-tab access. In supported streaming-platform sessions, navigating between titles while Blip is active can trigger additional summaries. Opening the chat can automatically start a summary; a separately typed question is not required for every request.

Data use and sharing

Blip uses the data described here to provide its summaries, questions and answers, content analysis, authorized Google Sheets reading, PDF reading and annotations, local transcription, read-aloud, editable dictation, local custom voices, history, and diagnostics. It does not sell user data, use it for targeted advertising, or use it to determine creditworthiness. Transfers to your selected AI provider and the service requests described above support these features. The developer does not receive your API keys, chats, transcripts, or cookies through a built-in collection endpoint. Information you voluntarily send through a support channel is received by that channel's operator.

Google API data use. Blip uses Google Sheets data to provide the user-facing reading, summary and question-answering features described above. Relevant excerpts are transferred to the AI provider you select for that purpose after you connect Google Sheets. The developer does not sell Google Sheets data or use it for advertising, creditworthiness decisions or a developer-operated model-training service. Google API data is subject to the Google API Services User Data Policy, including applicable Limited Use requirements. Blip's use and transfer of information received through Chrome APIs adheres to the Chrome Web Store User Data Policy, including its Limited Use requirements.

Children

Blip is not directed at children under 13. The developer does not knowingly collect personal information from children under 13. Contact the developer if you believe such information has been supplied through a support channel. Content sent directly to your selected AI provider is also subject to that provider's terms and privacy practices.

Changes and contact

Updates to this policy will be published at this privacy policy page with a revised effective date. The developer is an independent developer. For privacy or support questions, contact exergyoflife@gmail.com. Requests concerning data already sent to an AI provider must also be directed to that provider.