CleanApp for ChatGPT privacy policy
Effective and last updated: October 22, 2026.
This policy explains how CleanApp, the operator of the CleanApp ChatGPT miniapp and plugin, handles information through the tools at live.cleanapp.io/openai/mcp and the companion web preview. CleanApp is responsible for its processing described here. Contact us at b@cleanapp.io.
This is a complete notice for this integration. The general CleanApp privacy policy also describes other CleanApp products. If their descriptions differ, this specific policy governs this integration. ChatGPT and other AI hosts handle your conversation and their own records under their separate policies, including OpenAI’s Privacy Policy.
Preparing a preview already sends the selected report fields to CleanApp. It does not yet create a submitted backend report. Sending the approved preview creates a stored report for private review and analysis.
The current integration accepts text and optional coordinates. It requires no CleanApp account and has no photo or file upload tool. It does not automatically publish your report or send it to a responsible company or authority.
Draft tokens work for 24 hours and tracking tokens for 90 days. Those are validity periods, not deletion deadlines. Submitted backend records currently have no automatic expiry.
1. Information received and returned by each tool
When ChatGPT calls a CleanApp tool, it sends that tool’s arguments to CleanApp. We receive those selected arguments, rather than access to your entire ChatGPT account or conversation history. Your AI host can include conversation information in those arguments, so review what it proposes to send. All tool results below return to your AI host and may appear in your conversation or the miniapp.
| Tool | Information received by CleanApp | Information returned to ChatGPT or the web preview |
|---|---|---|
Open CleanAppcleanapp_open | No report fields. Technical request information still applies. | App name, intake availability, preview/publication status, listing link if configured, and an explanatory message. |
Find public reportscleanapp_search_reports | Search terms, physical/digital/all filter, language, and result limit. | Your query and matching public report titles, summaries, physical/digital classification, dates and CleanApp links, result count and explanatory message. Separate reporter identifiers and precise coordinate fields are omitted; public report text may still identify a place or contain personal information. |
Preview a reportcleanapp_prepare_report | Physical/digital domain; problem category; title; description; language; optional place description, latitude and longitude; and affected product or website for a digital issue. | The complete report preview; encrypted draft token containing the report and generated submission details; private tracking token; draft expiry; and explanatory message. The full preview and both tokens return to ChatGPT. Encryption does not hide the separately returned preview from your AI host. |
Send an approved reportcleanapp_submit_report | The draft token and explicit confirmation. We decrypt the token and send its approved report fields to the backend with generated identifiers, timestamps and integration provenance. | Receipt identifier, intake status, review lane, time received or updated, duplicate/retry indicator, explanatory message, private tracking token and tracking link. |
Check a receiptcleanapp_report_status | The private tracking token identifying that submission. | The receipt and intake information above. This tool does not return the full stored report or establish notification or resolution. |
The web preview sends equivalent fields directly to the integration. We do not request a ChatGPT password, contacts, payment information, government identifier, email address or automatic device location through these tools. Optional coordinates can be precise personal data; provide them only when you want that location included. A place description may reveal a home or workplace. We do not use device-location permission to find you.
Include only facts needed to describe the problem. Do not submit passwords, private keys, payment credentials, confidential records, government identification numbers, health information or other sensitive personal information. Avoid naming bystanders or including private contact details. Free text can contain personal information even without a dedicated field.
2. Other information we process
- Submission and analysis records: generated source, submission and receipt identifiers; creation and processing times; the integration’s agent identity and software version; confirmation/provenance flags; duplicate-prevention hashes; review lane/status; and analysis, classification, translation or moderation results. These identify a submission rather than a verified ChatGPT user identity.
- Technical and security information: hosting and web infrastructure process IP addresses, user-agent/browser information, request times, endpoints, response codes and related metadata to serve requests, limit abuse and investigate faults. An IP address may belong to your browser, AI host or an intermediary. Backend search requests include search terms in their request URL.
- Application logs: the integration’s own tool-call logs contain tool name, success/failure, error code where relevant and elapsed time. They do not intentionally record report text, draft tokens or tracking tokens. Infrastructure access logs are separate and can contain request metadata.
- Support correspondence: if you contact us, we receive your email address, message, receipt identifier and any supporting information you send.
3. Why we use information
We use search terms to find public observations. We use draft fields to validate input, show an exact preview and create encrypted tokens. We use approved submissions to receive and store reports, analyze and translate them, classify issues, assess privacy/redaction and moderation needs, preserve provenance, prevent duplicates and return receipts/status. Optional location or product details help interpret issues and identify relevant entities during review. Redaction flags do not guarantee that every identifying detail will be found or removed.
We use technical information to operate and secure the integration, apply rate limits, diagnose errors and prevent misuse. We use correspondence to provide support and handle corrections, complaints and privacy requests. Information may also be needed for legal obligations and legal rights.
We do not sell or rent private miniapp submissions, use them for advertising, build advertising profiles from them or use them to train CleanApp general-purpose AI models. AI analysis of a report is a separate processing step from general-purpose model training. We do not make a promise here about ChatGPT’s or another provider’s independent handling of records under its own terms and your settings.
Where applicable law requires a legal basis, we rely on providing the service you request, legitimate interests in operating a reliable and secure reporting service and preventing abuse, and compliance with legal obligations. We rely on consent where legally required and honor its withdrawal for that processing. Approving a report is also a product control over sending it; it does not waive privacy rights.
4. Who receives information
- Your AI host: OpenAI, when you use ChatGPT, receives the tool results above, including the full preview, public search results, tokens and receipt information. Your conversation may retain these after CleanApp tokens expire or CleanApp deletes its record. Manage host records separately.
- CleanApp personnel: authorized people operating the service, reviewing reports or handling support/privacy requests may access information needed for those tasks.
- Hosting and infrastructure providers: Google Cloud hosts the integration and backend infrastructure in the United States. Hosting, storage, networking and security providers process information needed to run and protect the service.
- AI analysis providers: submitted reports can undergo automated analysis during private review. The backend is currently configured to use Anthropic’s Claude API. Report text, relevant place/product context and generated analysis may be processed to classify, summarize, translate and assess the report. The backend also supports OpenAI and Google Gemini configurations; we will update this notice as appropriate if providers used for this integration change. Preparing an unsubmitted draft does not itself send it to the backend analysis pipeline.
- Location and contact research services: analysis of a physical report can send supplied coordinates or related place queries to OpenStreetMap’s Nominatim and Overpass services to understand the location. Where configured, Google Search can receive a place or organization name and city to find public contact information. CleanApp may fetch relevant public websites for that research. These lookups can occur during private review; they are distinct from sending the full report to a responsible party.
- Support and legal recipients: providers supporting correspondence may process your message. Professional advisers or authorities may receive information where necessary for legal obligations, a valid legal request, legal rights or service security.
We use providers for the purposes described here and seek appropriate contractual and security protections. Their processing and retention can also be subject to applicable provider terms. The integration does not send private report contents to advertising services.
5. Private review, publication and tracking links
A search reads existing public reports. A new miniapp submission enters CleanApp’s quarantine/private review lane. Private review includes storage and automated analysis; it does not mean that only the reporter can read it or that no provider processes it. This integration does not automatically publish submissions or notify a responsible company, property operator or authority. CleanApp has separate administrative review and promotion capabilities, so private intake is not a promise of permanent confidentiality. Any proposed further publication or routing must be evaluated separately under the applicable notice and authorization. A receipt is not proof of publication, notification or resolution.
A tracking link is a private access capability. Anyone obtaining it can read that submission’s receipt and intake status until expiry, without a CleanApp account or extra sign-in. Keep it out of public posts and shared screenshots. The token is placed after #track=: browsers do not send that fragment with the initial page request, but the web preview then sends the token to CleanApp to check status. ChatGPT receives it in tool results.
6. Retention and deletion
Token validity does not erase copies in a browser or AI conversation and does not determine a submitted report’s retention.
| Information | Current retention or validity | Your control |
|---|---|---|
| Unsubmitted draft | Processed to produce a preview/token, without a separate persistent draft database. Draft tokens are usable for 24 hours after preparation. Preview and token copies may remain in your browser or AI host longer. | Decline submission. Clear the web preview’s saved data and manage/delete the relevant AI conversation separately. |
| Tracking capability | Works for 90 days from preparation, even if the report is sent later. Expiry stops use of the token; it does not delete the report or copies of the token. | Keep it private. Contact us if exposed or if you need help with a stored report. |
| Submitted reports, receipts, provenance and analysis | No automatic expiry currently applies. Records can remain indefinitely until we act on a verified deletion request or determine they are no longer needed. Retention is assessed against review, status tracking, duplicate prevention, security, legal obligations and legal claims. Token expiry or app disconnection does not delete them. | Request access, correction or deletion at b@cleanapp.io, using your receipt identifier. Legal requirements or limited security/legal records may affect the response; we will explain relevant limitations. |
| Browser session data | The standalone preview saves the current draft and receipt in this tab’s session storage. It normally lasts for the tab session; browser session restoration can preserve it longer. Embedded mode does not use this feature; the host manages its own records. | Choose “Start another report,” clear this site’s browser data, or close the tab and disable session restoration as needed. |
| Operational/access logs | Integration application logs rotate at about 10 MiB per file, keeping up to three files. Web access logs rotate daily with 14 rotations. These settings are not a guarantee about every infrastructure/security record. Separate backend/security records have no single fixed retention period. | Ask about personal information in logs. We assess requests against operational, abuse-prevention and legal needs. |
| Backups and correspondence | Backups can contain older copies, including legacy backups; we do not promise a uniform fixed backup purge date. Correspondence is retained as needed to handle requests, preserve outcomes and meet security/legal needs, without automatic expiry. | A deletion response will explain scope, retained copies and exceptions. Deletion from active systems does not mean every historical backup is immediately erased. |
7. Your choices and privacy rights
You can search without submitting, omit optional coordinates, edit a draft before preparing a replacement, decline submission, stop using the integration, or disconnect it through your AI host’s settings. The standalone preview uses session storage for drafts and receipts; it does not use advertising cookies or an analytics identifier. “Start another report” clears those saved entries. Clearing browser data or disconnecting does not delete submitted records or your host’s conversation.
Depending on your location and applicable law, you may have rights to access or obtain a copy of personal information, correct it, request deletion, restrict processing, object, receive portable data, withdraw consent where processing relies on it, and complain to your privacy/data-protection authority. These rights have legal conditions and exceptions. We assess requests and respond within applicable legal periods without discrimination for exercising a right.
Send requests to b@cleanapp.io. State your request and include the receipt identifier and enough context to find the record. Without a receipt, describe the report and approximate submission time. Because no account is required, we may need proportionate verification that you are entitled to access or change information. Do not email private tracking tokens, passwords, government ID documents or other secrets. We will discuss a suitable verification method if needed. We can address personal information about you even if another person submitted it.
Use ChatGPT’s or your other AI host’s controls and privacy-request process for information it holds independently. A request to CleanApp does not automatically reach providers acting in that independent role.
8. Security and international processing
We use HTTPS connections, authenticated backend service access, encrypted and integrity-protected draft/tracking tokens, input validation, request limits and restricted operational access. Safeguards reduce risk but cannot guarantee absolute security. A token grants its associated access capability; encryption does not make sharing it safe.
Our Google Cloud infrastructure is in the United States. AI and other providers may process information in other countries with different data-protection laws. Where applicable law requires safeguards for an international transfer, we use the required lawful transfer mechanism and contractual safeguards. Contact us for information about safeguards applicable to your information; this notice does not claim every transfer is covered by a particular certification or agreement.
9. Children and policy changes
This integration is not directed to children under 13. You must also meet your AI host’s and local law’s minimum-age and parental-permission requirements. We do not knowingly solicit personal information from children under 13. Contact b@cleanapp.io if you believe such information was submitted so we can investigate and take appropriate action.
We date changes on this page. For material changes affecting how we use or share information, we provide additional notice or obtain consent where legally required. Send questions, complaints and privacy requests to b@cleanapp.io.