Privacy Policy
Last updated: 2025-12-11
Arc ("we", "us", "our") is operated by Good Morning Technologies GmbH, a GmbH registered in Germany. This policy explains how we handle personal data when you visit arc.so, use our creation, storage, publishing, audience and commerce services, use our desktop tools, or contact us (together, the "Services"). It also explains your choices and rights.
We process account details, content and technical data to provide the Services. Additional data is shared when you use features such as AI assistance, newsletters, payments or connected applications, as explained below. Contact [email protected] about your data.
1. Who is responsible for your data
The controller responsible for our account, service-operation and support processing is:
Good Morning Technologies GmbH
Legal form: GmbH
Registered address: Linienstr. 41, 10119 Berlin
Country: Germany
Registration details: Registergericht: Berlin (Charlottenburg) - HRB 271522 B. Sitz der Gesellschaft: Berlin
Legal representative: Dmitry Monin
Privacy contact: [email protected]
{{dataProtectionOfficerOrRepresentativeDetails}}
Creators may decide how personal data is used in their publications, subscriber lists and customer relationships. Where we process that data on a creator's behalf and instructions, the creator is the controller and we act as processor under our separate Data Processing Agreement. We remain controller for our own account administration, billing, security and support purposes. The role depends on the particular activity; operating a publication does not by itself make every use of data processor activity. If your request concerns a creator's publication, you can contact the creator or contact us for assistance.
2. Personal data we process
- Account and connection data: name and email where supplied, username, account identifiers, sign-in information, account settings, connected-client identifiers, permissions, and authorization/revocation records. Our authentication service processes the credentials needed for sign-in.
- Photos and other content: uploaded files and processed copies, filenames, dimensions, captions, story text, drafts, revisions, selection/order information, publication settings, destination URLs and publishing records. Content can identify you or other people depicted or mentioned.
- Embedded photo metadata: where present, camera/lens information, capture date and time, descriptive tags, and GPS coordinates or place information. We extract metadata for media organisation and related functions. Remove metadata you do not want us to receive before uploading. Publication and metadata controls are described in Section 8.
- Communications and audience data: information you send in support requests, feedback or other correspondence; subscriber names and email addresses, subscription and unsubscribe preferences; message content and delivery, bounce or complaint records. Where email open/link tracking is enabled, related activity records may also be processed, as described in Section 10.
- Commerce and reminder data: billing and order details, payment and subscription references, buyer and recipient contact information, delivery addresses, and phone numbers supplied for text reminders or verification, when you use these features.
- AI feature data: the text, selected image copies, contextual metadata and instructions described in Section 4, generated results, and records of the operation and associated usage/credits.
- Technical and usage data: IP address, browser/user-agent and connection information, request times, errors, service actions and security/rate-limit records as needed for delivery, security and troubleshooting. Section 10 explains first-party analytics.
- Other enabled service data: {{otherServiceDataCategoriesAndPurposes}}.
Technical identifiers and pseudonymous records may still be personal data. We describe statistics as anonymous only when people cannot reasonably be identified from them. A hashed identifier is not automatically anonymous.
Please upload only content you are entitled to share. Our creative AI features are not intended for passwords, payment-card details, government identification documents or protected health information. Avoid including these in photographs, captions, story text or AI instructions.
3. How we obtain data
We receive data directly when you register, upload files, edit or publish content, use account controls or contact us. Technical data also arises when your browser or connected client requests our services. Connected services provide the information needed for the functions you authorize. Other people may provide information about you in their uploaded content or communications.
Account and content information necessary to deliver a requested function must be provided for that function to work. Tools such as annotation, proofreading and cover generation, and connections such as ChatGPT, are optional. Other processing can run automatically within a feature you use, including publication translations and service checks described in Section 4. Cookie and analytics details are in Section 10.
4. AI assistance and optional connected applications
4.1 AI features within the Services
When you request an AI tool or use a function that includes AI processing, we send the relevant inputs to the provider for that function. A request may include unpublished material and personal information about people depicted or mentioned in it. These features use provider APIs through our Services and do not require a ChatGPT account or connection.
| Feature | Provider | Information and purpose |
|---|---|---|
| Photo annotation, including the desktop annotation tool | OpenAI | A contact sheet of selected photos, filenames, supplied capture dates and location labels, context, credits and language/style instructions to generate captions, tags and descriptive information. |
| Caption rewriting | OpenAI | The full caption, any selected excerpt, supplied context and language instructions to improve, shorten, simplify or expand the wording. |
| Story proofreading | OpenAI | The story title and text from its headings, articles and text blocks, including relevant nested or introductory text. This can include the full text of an unpublished draft, rather than just a selected sentence. |
| Story translation | OpenAI | Story text segments and source/target language instructions to produce translated versions. Publishing with an eligible active creator plan can automatically trigger translations into subscribers' preferred languages; a separate translation request is not required for each publication. |
| Cover analysis and generation | OpenAI for analysis; Google Gemini for generation | OpenAI receives a reference image copy, story title, text and facts, design preferences, and creator/publication names where selected. Google receives the resulting or edited generation prompt, visual identity instructions and any selected reference image copy to generate the cover. Prompts can contain information derived from the story. Generation may use Google Search to inform the result. |
| Instagram carousel planning | OpenAI | Story text and related image identifiers, captions and descriptive metadata to suggest slide text and arrangements. This planning request uses text and metadata rather than image files; planning does not itself publish to Instagram. |
| Publishing schedule interpretation | OpenAI | Your written schedule instructions, timezone and relevant date to propose a publishing schedule. |
For product mockups, Google Gemini is used to prepare background scenes from scene instructions and any reference photograph supplied for that preparation. When you choose an existing background for your product, our server composites your artwork onto it without sending the artwork to Gemini for that operation.
4.2 Automated checks and enabled reminders
Where configured, we also use OpenAI to check a proposed username for reserved terms, inappropriate wording or impersonation risk. This check runs when a username is set or changed and can cause that username to be refused; it is separate from optional creative AI tools. Contact [email protected] if you need assistance with a rejected username.
For enabled publishing reminders, OpenAI may receive your first name, publication title and publishing schedule/activity facts to help write reminder messages. Once reminders are enabled, these requests can run automatically when a reminder is due. The email or SMS delivery provider then receives the reminder content and recipient address or phone number.
4.3 Provider handling and our AI records
Provider processing and retention depend on the relevant API service, contract and settings. OpenAI's published API policy states that API inputs and outputs are not used to train its models by default unless sharing is enabled. Google's Gemini terms distinguish data-use arrangements by service, billing and region, and include additional handling for Google Search grounding. These API arrangements differ from a user's own ChatGPT account. Our applicable provider arrangements, retention and controls are: {{aiProviderDataUseAndRetentionDetails}}.
We retain generated content where needed to deliver the feature, and operation records for retries, usage/credit accounting and troubleshooting. Depending on the feature, these records can include prompts, input text or excerpts, outputs and metadata. Proofreading omits the input story from its durable prompt log, but saved suggestions can quote story text. Disabling provider response storage does not itself remove our records or eliminate provider safety/security retention. Our retention schedules are in Section 9.
AI results can be inaccurate. Review descriptions, translations, suggestions and the intended audience before saving, using or publishing them.
4.4 For users using the ChatGPT App (optional)
If you choose to connect your Arc account in ChatGPT, you sign in through our authorization flow and grant the displayed permissions. Depending on that grant, the connection can read eligible images and stories, upload photos, save drafts and publish confirmed stories. We retain authorization records and scoped tokens to operate and secure the connection. You do not enter your Arc password in the ChatGPT component, and OAuth tokens are not included in model-visible tool results.
We receive task-specific tool requests, including image/story/destination identifiers, titles, captions, story text, ordering instructions, revisions and publication confirmations. We do not request your complete ChatGPT conversation history. What you tell ChatGPT can reach us when included in a tool request or saved into a story. If you use ChatGPT's voice input, ChatGPT handles it under its own terms; our story tools do not receive raw audio.
Our tools return destination names, image identifiers, filenames, captions, dimensions and preview links; story text, order, revisions, visibility and destination details; and upload/publication status. Browsing the library can return metadata and preview links for a page of available photos before you select pictures. Image inspection returns reduced-size copies of selected photos. The component also shares current selections and staged draft text with the conversation, including unsaved edits. Device uploads go to our authorized storage destination; subsequent sharing of previews and content with ChatGPT is a separate exchange.
OpenAI provides ChatGPT separately. Information it receives is subject to the terms, privacy information, retention and model-improvement controls applicable to your ChatGPT account or workspace. Saving a draft does not publish it. Disconnecting does not erase content already saved in either service; deletion must be requested separately where appropriate. Section 8 explains connection and publication controls.
5. Purposes and legal bases
Where EU/EEA or UK data-protection law applies, the basis depends on the processing activity:
- Providing the requested service: account management, storage, editing, requested AI assistance, commerce and user-requested publication are processed as necessary to perform our contract with you, where that basis applies to the particular data and activity. Optional connections enable the functions you authorize.
- Security and support: preventing misuse, checking usernames, controlling access, maintaining reliability and handling service enquiries support our legitimate interests in operating and protecting the Services, subject to the required balancing against your rights.
- Analytics and improvement: {{analyticsLegalBasisAndControls}}.
- Consent: where required for non-essential cookies, marketing or another optional use, we request consent separately. You may withdraw it without affecting earlier lawful processing.
- Legal obligations: we process records needed to meet applicable legal duties, including relevant accounting and lawful disclosure obligations.
Granting a technical connection permission is not blanket legal consent for unrelated uses. We assess the appropriate basis separately for information concerning other people in creator content. {{otherProcessingLegalBases}}
6. Who receives data
- Hosting and storage: Hetzner and Google Cloud provide hosting, storage, infrastructure and media delivery under the relevant service arrangements.
- AI providers: OpenAI and Google Gemini receive the inputs for the features and checks described in Section 4. These are distinct from using the optional ChatGPT connection.
- Email: Postmark, or the configured SMTP provider, receives sender/recipient details, message subjects, bodies and relevant headers/links to deliver sign-in, transactional, support and newsletter emails. Sending a story as a newsletter shares the content included in that email with the delivery provider. Postmark also processes inbound support emails where enabled, including attachments supplied to that email service, even if our support inbox does not store those attachments. Delivery, bounce, complaint, suppression and enabled open/link activity records may be returned to us.
- Payments: Stripe processes checkout, subscriptions, billing and payouts where used, receiving relevant customer, transaction and order details. Stripe also has its own obligations and privacy information for processing it performs as a separate controller.
- Optional text and address features: Twilio receives phone numbers and SMS or verification data for enabled text features. Google Places receives entered search/address text, selected place identifiers and session tokens for address suggestions/details where used.
- Fonts and icons: pages loading fonts or icons from Google Fonts make requests directly from your browser to Google. Those requests disclose your IP address and browser/request information to Google, even if you do not use an AI feature or connected application.
- Printing and delivery: MyPostcard receives the print-ready postcard, photograph and recipient name/address when you order a physical postcard. Sellers and fulfilment recipients receive the order/contact details needed to fulfil your purchase.
- Diagnostics: Sentry, where configured, receives error and operational diagnostic data to help investigate failures. Such data can include request context, identifiers or content appearing in errors, depending on the event and controls applied.
- Connected services: if you use the ChatGPT connection, OpenAI receives the information described in Section 4.4. Other connected services receive data relevant to the functions you authorize; choosing an external sign-in provider also involves that provider's authentication processing.
- Your chosen audience: published content is made available according to its access settings. Public content can be viewed, copied, shared or indexed by people and services outside our control.
- Authorized personnel and professional advisers: where needed to operate the Services, respond to you, protect legal rights or meet obligations, with appropriate access restrictions.
- Authorities and other legally required recipients: where disclosure is required by law or necessary and lawful to address fraud, misuse or threats to rights and safety.
The applicable provider legal entities, enabled services, roles and any additional recipients are: {{serviceProviderDetails}}. Providers acting on our behalf are subject to appropriate contractual obligations. Services acting as separate controllers have their own privacy information. We do not sell or rent your personal data. {{additionalDataSharingDisclosures}}
7. Storage locations, transfers and security
Our primary hosting and storage locations are {{primaryHostingAndStorageLocations}}. Processing by AI, email, payment, diagnostics, fulfilment and connected-service providers may involve locations outside the EU/EEA or UK. Hosting our own application in the EU does not mean every recipient processes data only in the EU.
For transfers we are responsible for, the applicable destinations and safeguards are {{internationalTransferDetails}}. Where required, these may include an applicable adequacy decision or contractual safeguards appropriate to the transfer. You may contact [email protected] for information about, or a copy of, the relevant safeguards, subject to appropriate redactions.
We use technical and organisational measures appropriate to the risks, including access restrictions and controls over account authorization, private media delivery and publication. No system is completely secure. This does not limit our obligations under applicable data-protection law.
8. Your content, publication and connection controls
You can review content and its audience before publication. For the optional ChatGPT connection, saving or previewing does not publish the story. Its publication review identifies the destination and audience and indicates when the destination space will also be published. Publishing through that connection does not send a newsletter. Newsletter sending is a separate function and shares the email content with the delivery provider and its recipients.
Photo metadata and the controls governing its public display are: {{photoMetadataVisibilityAndControls}}. Information visible in an image or its text can disclose a location even if embedded GPS metadata is removed.
You can manage saved content through the available Arc account controls or contact us for help. Removing a photo from a draft does not itself erase that photo from your media library or from an already published version. Public copies made by others may remain after you remove content from our Services.
To stop using the connection, disconnect the app in ChatGPT. To request revocation of outstanding Arc authorizations, contact [email protected] or use any connection-revocation controls available in your account. Revocation does not erase saved photos, drafts, published stories, connection-history records or content already received by ChatGPT. Request deletion separately in each service as appropriate.
9. How long we retain data
We retain data for the purposes described above and apply the schedules below. A lawful preservation requirement may extend retention for specific records; access and use remain limited to that purpose. Token or link expiry limits access and is not equivalent to deletion of the underlying content.
| Category | Period or criteria |
|---|---|
| Account data, saved media and stories | Kept to provide your account and stored content until deletion is requested or another disclosed account-closure rule applies. Active-system deletion and any exceptions: {{accountAndContentDeletionSchedule}}. |
| Temporary uploads and processing records | Raw uploads are normally removed after processing. Abandoned/failed uploads and processing records follow {{temporaryUploadRetention}}. |
| Connection, token and permission history | {{connectionRecordRetention}}. Revoked or expired connection records can remain for replay prevention and security until their scheduled removal. |
| Publishing receipts and action history | {{publicationAndActionRetention}}. These records support reliable retries and the history of requested changes. |
| Security and operational logs | {{securityAndOperationalLogRetention}}. |
| Support and email records | {{supportAndEmailRetention}}. |
| AI operation records and generated results | {{aiOperationRetentionSchedule}}. This covers stored prompts/input excerpts, outputs, retry records and usage/credit history. Results saved as media or story content also follow the content schedule. Provider-side retention is described in Section 4.3. |
| Analytics and activity records | {{analyticsRetentionSchedule}}. |
| Accounting and other legally required records | {{statutoryRecordRetention}}. |
| Backups | {{backupRetentionAndDeletionSchedule}}. Deleted records may remain in restricted backups until scheduled expiry; restored data is subject to the applicable deletion controls. |
When information is no longer needed, we delete it or make it genuinely anonymous. Account deletion does not necessarily remove every record immediately: backups, required records and copies held by separate recipients follow their applicable schedules. ChatGPT's retention and deletion controls apply separately to information it has received.
10. Cookies, analytics and third-party content
We use necessary session and security technologies for functions such as sign-in. The current inventory, purposes, durations and available controls are {{cookieInventoryAndControls}}. If non-essential cookies or similar technologies require consent, we obtain it before using them. The absence of advertising cookies does not itself mean there is no personal-data processing.
Our first-party publication analytics counts eligible page/story views and their source category. To estimate daily visitors, it can derive an account-and-day-specific keyed identifier from IP address and user-agent information. The analytics pipeline discards those raw inputs after deriving the identifier; this does not describe separate infrastructure/security logs. Creators receive summary counts rather than the raw inputs. We treat linkable identifiers as personal data.
Other usage measurement, including enabled creator activity or editing-time measurement, and your controls are: {{additionalUsageMeasurementAndControls}}. Retention is set out in Section 9. Browser settings let you manage stored cookies, although disabling necessary cookies can prevent sign-in or other functions.
Email delivery records are used to send messages and manage failed delivery, complaints and unsubscribes. Where email open/link tracking is enabled, the delivery provider can record when a message is opened or a tracked link is followed and associated technical information. Our enabled tracking, purposes, legal basis and controls are: {{emailTrackingDetailsAndControls}}.
Third-party links and embedded services have their own privacy information. ChatGPT and its interface may use their own cookies and technologies. We are responsible for our own processing and integrations; this policy does not replace the notices of separate services you choose to use.
11. Your privacy rights
Depending on applicable law and its conditions, you may request access to your personal data, correction, erasure, restriction, portability, or object to processing. You may withdraw consent for processing based on consent, and object to direct marketing. You can also complain to the appropriate data-protection authority.
Contact [email protected] to exercise these rights. We may ask for proportionate information to verify your identity. We respond within applicable legal deadlines and explain any lawful limitation. You can also use available account controls to manage your content. For data held separately by ChatGPT, use its privacy controls or contact its provider.
Deleting our account data does not automatically delete another person's lawful records or copies held independently by others. We assess deletion requests and any legal exceptions rather than promising immediate removal of every copy.
12. Children and other people in photographs
The Services are not directed to children under 13. Applicable age requirements and any parental-authorization rules depend on local law. Our account eligibility and parental process are {{ageEligibilityAndParentalProcess}}.
If you believe a child has supplied personal data contrary to these rules, contact [email protected] so we can investigate and take appropriate steps. Photographs uploaded by adults can also contain children's personal data. Consider the child's privacy and obtain any permission required before uploading or publishing such content.
13. Changes and contact
We update this policy when our practices or applicable requirements change and identify the current version by the date above. We provide notice of material changes where required. If a new use requires your consent, we request it separately; continued use is not treated as explicit consent to that new use.
For questions or requests, contact Good Morning Technologies GmbH, Linienstr. 41, 10119 Berlin, Germany, at [email protected]. Website: arc.so.