MingleLab · Public information

Privacy Policy

1. Who operates MingleLab

MingleLab is operated by MingleLab. This policy applies to the MingleLab website and application. For privacy questions or requests, email support@minglelab.co.in.

The service is for adults aged 18 or older. Please do not register or submit information about children. Contact us if you believe a child has created an account.

2. Information we collect

Public website usage insights: public information pages can send daily aggregate page-view, navigation-click and video-play events, plus loading/layout measurements. These insights store page paths, broad device/referrer categories and counts, not account identifiers, IP addresses, raw referring URLs, query strings or conversation content. They are not unique-person counts. You can turn this reporting off using the public footer; browser Do Not Track and Global Privacy Control are respected. Aggregate records are retained for up to 180 days. Infrastructure and security logs are handled separately as described in this policy.

Optional agent check-ins: you may separately enable messages, audio-call invitations and video-call invitations from fictional partners you already chat with. We store your choices, invitation identifiers, timestamps, delivery attempts and acceptance/decline state. Check-in text is selected locally using bounded conversation context and the fictional partner's mood; it is saved in the normal chat. Invitations never open your microphone/camera or start a paid AI call automatically. Started calls and your replies use the normal AI-credit rules. Generic browser notifications contain no chat text. Quiet hours, daily limits, cooldowns and unsubscribe controls apply; browser delivery is not guaranteed.

Optional AI voice calls: after you start a call and grant microphone permission, short recordings are sent to OpenAI for transcription. Sent voice transcripts, selected language, call identifiers, timestamps and partner replies are stored in the same conversation history as text chats, including administrator transcript access and export. You may review a transcript before sending it. MingleLab does not persist raw microphone recordings or generated playback audio; providers process audio under their own applicable terms and retention controls. Voices are synthetic, not real people. Speech and translations may be inaccurate. Existing separate chat-training consent applies to saved voice transcripts too; starting a call does not grant training consent.

Optional video calls are AI simulations, not live human video. The original mode blends prepared fictional facial poses in your browser for approximate lip sync. Enhanced live video, where enabled, asks for your agreement before sending the fictional portrait and the AI partner’s generated speech to Protoface to render synchronized video. Your microphone goes to the voice provider as described above, not to Protoface. Provider processing and retention are subject to the provider’s service terms. Enhanced video uses additional credits and is limited by the provider plan. Your optional camera preview is processed locally, not uploaded, recorded or sent to the AI partner. Turning it off, closing the call or hiding the page stops camera capture. Voice transcripts and credit accounting work as described above.

Account and profile information: username, display name, email if supplied, declared age, gender, partner preferences, biography, interests, languages, profile photos and any coach application details you choose to provide. Password-based accounts store a salted password hash, not a readable password.

Conversation and learning information: saved AI-coaching questions, answers and editable suggestions, human-coach context-sharing choices, message delivery acknowledgements, optional read receipts, short-lived typing signals, likes, passes, matches, messages, quoted replies, conversation states, AI drafts, translations, coaching feedback, practice sessions, saved insights and reflections. Conversations with simulated agents are fictional; conversations with real members or coaches can contain personal information.

Product feedback: feature, submission type, title, description, optional star rating, status, priority, timestamps and comments exchanged with the MingleLab team. This feedback is visible only to its submitting account and authorised administrators, not other members. We use it for support and product improvement; it is not automatically used for model training. Do not include passwords, payment credentials or unnecessary sensitive information. Feedback and comments are retained with account/support records; contact support for privacy or deletion requests.

Optional Virtual Life stories: fictional scene choices, story settings, simulated relationship stages, game meters, skill progress and a timestamped outcome history are saved with your account and can be reviewed by administrators. Stories combine a lightweight local game engine with AI-generated partner reactions and communication insights, not predictions about real people. Relevant profile details, selected choices, recent conversation excerpts and recent story context are sent to the configured AI provider. AI reactions deduct credits according to actual token usage; navigation and saved history do not. Dramatic betrayal storylines require a separate opt-in, and you can pause or leave without losing progress. These fictional outcomes do not change your real relationships or your saved chat's connection state. Story choices are not automatically used to train a model; any future research use is subject to your research consent and the safeguards in this policy.

Optional scam warning-sign practice is off by default. If enabled in Profile, some fictional agents can present labelled, hypothetical fraud-pressure exercises based on public advisories. We save the exercise pattern, source references, message evidence, observed response and any standardized report. A coach explanation can be requested without spending credits. Turning practice off stops future exercises; it does not delete saved records.

Usage and security information: sign-in attempts, session identifiers, pages and features used, timestamps, approximate device/browser categories, errors, request timing, credit usage and payment records. Infrastructure providers may also process network and IP information for delivery and security. Avoid sending financial credentials, identity documents or unnecessary sensitive details in chats.

3. Google sign-in

If you use Google sign-in, Google supplies a verified account identifier, email address, display name and profile-picture URL. We use these to create or recognise your MingleLab account and display your default avatar until you replace it. We do not request access to your Gmail messages, contacts, Google Drive files or Google password.

Your displayed name and avatar can be visible as part of your member profile or to administrators. Choosing Google sign-in does not make your profile automatically discoverable. You can revoke the Google connection in your Google Account settings; revoking it does not by itself delete information already held in MingleLab.

4. Why we use information and who can see it

We record the version and time of the acknowledgement made during new self-service signup. Safety reports contain your description and a reference to the saved conversation. Administrators review reports; the service is not monitored continuously. Blocking ends further contact and hides member discovery profiles from each other, but does not erase saved transcripts or recall previously downloaded photos.

We use information to authenticate you, maintain your profile, suggest matches, deliver chats and coaching, track learning progress, manage credits and payments, prevent abuse and diagnose failures.

When you enable real-member discovery, other signed-in members can see the profile information offered for discovery. Chat participants can see their shared conversation. Your email, password hash and private partner-preference settings are not included in member discovery profiles.

The optional AI-coaching panel uses the selected conversation and your coaching questions to offer guidance and editable drafts. A human coach does not automatically see your other conversations. You can explicitly share a selected agent or real-member conversation with a connected coach; access shows only the latest permitted excerpt. Turn sharing off to stop future access. Information the coach already read or copied cannot be recalled. Do not share passwords, payment credentials or another person's unnecessary private information.

Authorised administrators can review account activity and full conversation transcripts, including agent-to-member, member-to-member and coaching chats, and can download transcripts for support, safety and operational review. Chats are not end-to-end encrypted and should not be treated as confidential from the application operator.

5. AI, hosting and payments

Google Cloud hosts the cloud application, database, stored photos and operational logs. OpenAI processes the relevant conversation context, profile information or image prompts needed for enabled AI replies, drafts, translation, coaching, matching and fictional images. Only use these features if you are comfortable with the required information being processed by these providers. Their own terms and privacy practices also apply; processing may occur outside your country.

Razorpay processes payments when enabled. MingleLab stores order/payment references, purchased credits and transaction status, rather than collecting your full card details or UPI PIN. Google sign-in and Razorpay checkout may use their own cookies or similar technologies.

We do not sell personal information or use the app's activity records for targeted advertising. Information may be disclosed when necessary to respond to lawful requests, protect people or investigate abuse.

6. Optional research and future model development

The optional research setting is off by default and can be changed in You → Profile. When enabled, activity records may be included in research-mode exports without names or message text. Withdrawing the setting excludes your historical activity from subsequent research-mode exports; it does not erase operational records or copies already exported.

This setting is not permission to train models on your chat text, photos or Google account information. A separate, optional chat-training permission in You → Profile allows pseudonymised conversation excerpts to be reviewed by authorised data reviewers and used in scam-detector training and evaluation. Every real participant must give that separate permission before a conversation can be imported. Reviewers check for identifying details, but automated removal may not catch every detail. Do not include secrets or unnecessary personal information in research conversations.

The research pipeline versions datasets and trained detectors. Withdrawing chat-training permission blocks affected excerpts from subsequent training, export and deployment use; it cannot recall earlier downloads or instantly erase information from previously trained model weights. Contact support about deletion and remediation requests. Administrators' operational transcript access and live safety analysis are separate from optional research participation. Detector scores are uncertain warning signals, not proof of fraud, and are not used for automatic account bans.

7. Cookies, security and retention

MingleLab uses a session cookie and browser storage needed for sign-in, security and the interface. Optional learning reminders are off by default. Browser push separately requires your permission and subscription; browser push services receive the endpoint and encrypted generic notification, not your chat text. Reminders follow your saved schedule and quiet hours, are not marketing consent, and can be unsubscribed in You. Delivery depends on browser support and is not guaranteed. Email reminders are not enabled. The app does not add third-party advertising trackers.

We use access controls, password hashing and secure transport on the deployed HTTPS service. No service can promise complete security. Keep your credentials private and contact us about suspicious activity.

Account details, photos, transcripts and learning records are retained while needed to operate the account; there is no automatic account-wide deletion deadline while an account remains active. When you request account deletion, account access closes and optional research permissions are withdrawn. A durable worker waits for in-flight operations to drain, then deletes profile photos, operational account data, learning history and shared conversation threads, including their copies in other participants' chat history. Temporary service failures are retried. Minimal payment, refund, payout and safety records needed for legal obligations or disputes may be retained separately. Activity audit retention defaults to 180 days for active accounts. Backups expire under the configured backup retention policy and previously downloaded exports cannot be recalled. Opaque deletion tombstones prevent stale workers from recreating deleted operational records.

8. Your choices and requests

You can edit profile details, replace or remove uploaded photos, change discovery and research settings, close conversations and sign out. To request permanent account deletion, sign in, open You and choose Delete my account. Recent sign-in is required; you may be asked to sign out and sign in again first. A deletion receipt can be used to check progress. For access, correction, privacy complaints or help with deletion, email support@minglelab.co.in. Never email a password. Visit /account-deletion for instructions outside the app.

Applicable privacy rights depend on your location. We may update this policy as the service changes and will show the revised date here. Material new data uses require appropriate notice and, where needed, consent before they begin.