Botticelli Privacy Policy

1. About Botticelli

Botticelli is a Discord application providing moderation, community management, member onboarding, participation tracking, community challenges, and related services in participating Discord servers.

This policy covers the Botticelli application and the information it processes. It does not cover other Discord applications.

Botticelli is maintained by its development and administration team.

Some components are being migrated from an established production system to a newer multi-server architecture. The newer architecture introduces additional privacy protections, configurable access controls, and more selective data collection.

During this transition, legacy and newer components may operate alongside one another.

2. Information We Process

Depending on the features enabled in a participating server, Botticelli may process the following information.

Discord account and membership information

• Discord user IDs.
• Usernames, display names, and mentions.
• Server membership and join dates.
• Server roles and relevant permissions.
• Onboarding status.
• Member participation and administrative records.
• SHA-256 fingerprints of user and server avatars, used to detect staff impersonation and duplicate avatars.

Messages and interactions

• Message text and associated metadata.
• Message edits and deletions.
• Channel, message, and server identifiers.
• Reactions and button interactions.
• Commands and related responses.
• Images and other attachments when relevant to moderation or community features.

Community participation

• Participation in community challenges and voting.
• Experience points and rank progression.
• Tokens, balances, and transactions.
• Activity and engagement statistics.

Moderation information

• Suspected violations of server rules.
• Automated moderation results.
• Moderation actions and related records.
• Information needed to investigate or review moderation decisions.
• Limited evidence associated with privacy or safety incidents.

Not every feature collects every category of information.

3. Why We Process Information

Botticelli uses information to:

• Detect and respond to prohibited content, spam, scams, and privacy violations.
• Assist authorized moderators.
• Manage onboarding, roles, and member status.
• Operate community challenges, voting, and rewards.
• Maintain participation statistics.
• Track experience points and tokens.
• Respond to commands and provide community assistance.
• Maintain system security and investigate technical problems.

Botticelli does not sell Discord member data or provide it to advertising networks or data brokers.

Why Botticelli Requires Privileged Discord Intents

Botticelli requires two privileged Discord intents to provide its community-management and moderation functions.

Message Content Intent

Message Content is used for:

• Automated moderation scanning.
• Statistical collection in configured server locations, subject to server-specific exclusions.
• Daily challenge entries and voting.
• Recording activity dates used in inactivity detection.

These features require access to ordinary server messages, including messages that do not mention Botticelli or invoke an application command.

Server Members Intent

Server Members is used for:

• Membership sweeps and membership-status reconciliation.
• Member onboarding and preboarding.
• Inactivity detection and related role management.
• Rank and experience-point role reconciliation.
• Impersonation checks involving member identities and profiles.

These functions require membership events and current membership or role information.

4. Message Collection

Some Botticelli features require access to messages that are not directed specifically at the application.

These include automated moderation, participation tracking, and certain community activities.

Botticelli's newer statistics architecture allows server administrators to exclude specific channels, categories, threads, or messages from statistical collection.

Messages created or edited in excluded locations are not captured by the new statistics system. These exclusions do not disable moderation scanning.

When a message is deleted, the newer system removes its stored text and edit history from statistical records, including when the deletion occurs in an excluded location.

These controls are being introduced as part of the ongoing migration. Legacy components may continue to operate during the transition and may have different storage and retention behaviour.

5. Privacy and Anti-Doxxing Protection

Botticelli includes automated safeguards intended to identify and remove certain forms of sensitive personal information that should not be publicly disclosed.

Depending on the context, this may include private contact information, financial information, or specific personal addresses.

The newer moderation architecture is designed to avoid unnecessarily reproducing sensitive information in public explanations, ordinary staff notifications, or statistical records.

Where limited evidence must be retained for review or appeals, the newer system uses masking, restricted access, and additional auditing safeguards.

These safeguards are being rolled out as part of the migration.

6. Automated Analysis and AI Services

Certain Botticelli features use automated analysis and third-party AI services, including the OpenAI API.

The information submitted depends on the feature being used.

Moderation and Profile Screening

For moderation evaluations, information sent to OpenAI may include:

• Message text.
• Reporter notes.
• Images or links to them.
• Applicable server rules.
• Moderation instructions.

When members join or update their profiles, Botticelli may also submit names, nicknames, and avatar images to OpenAI to check for impersonation and potential policy violations.

These requests do not create persistent conversation records through the API's response-storage features.

Welcome Messages

For optional welcome-message generation, Botticelli may submit answers provided through an introduction form, such as a member's profession, interests, or intended use of the community's tools.

These requests do not create persistent conversation records through the API's response-storage features.

AI-Assisted Commands

Botticelli provides AI-assisted features such as ideation and knowledge assistance.

Depending on the feature, member questions, supplied content, attached images, and relevant conversational context may be submitted to OpenAI.

Some commands process individual requests without storing persistent conversation records.

Knowledge Assistant and Conversation Storage

When enabled, Botticelli's Knowledge Assistant can respond to members who mention the bot or reply to its messages in channels the bot can access, subject to configured exclusions. It can also respond automatically in designated help channels.

To understand questions and provide relevant answers, the Knowledge Assistant may process:

• The member's question and attached images.
• The message being replied to, including messages written by other members.
• Replies from community volunteers, including their display names.
• Surrounding channel messages from other members, including display names, when contextual message collection is enabled.

This information, together with generated responses, may be stored by OpenAI as persistent Response objects to maintain conversational context.

Botticelli temporarily retains the associated response identifier in Redis to continue the conversation.

The Knowledge Assistant may also retrieve information from a separate OpenAI-hosted vector store containing the application's knowledge-base documents.

Retrieval from this knowledge base does not add members' messages or conversations to the vector store.

Earlier versions of some Botticelli features used OpenAI's Assistants API threads. Those features are no longer active.

Data Retention by OpenAI

Moderation, profile screening, welcome-message generation, and simple assistance requests do not use persistent response storage.

When enabled, the Knowledge Assistant uses stored OpenAI Responses to maintain conversational context.

OpenAI's standard Responses API retention period is 30 days. Its documentation states that stored Responses are retained for at least 30 days under the default settings.

Botticelli does not currently delete these stored Responses earlier. Expiration of the locally retained response identifier does not delete the corresponding data held by OpenAI.

OpenAI may separately retain API inputs and outputs for abuse monitoring and other purposes under its applicable policies.

Botticelli does not use members' messages to train or fine-tune its own AI models.

7. Storage and Access

Botticelli uses databases and supporting infrastructure to operate its features.

Access to stored information is limited according to operational responsibilities.

The newer architecture includes additional protections such as:

• Server-specific statistical collection settings.
• Pseudonymous identifiers for certain records.
• Restrictions on access to sensitive moderation evidence.
• Auditing of sensitive evidence access.
• Configurable authorization for staff commands.

In the newer architecture, staff and moderation commands require Manage Server permission by default, although server administrators may authorize additional roles.

Pseudonymization reduces direct identification but does not make all information anonymous.

8. Data Retention

Retention depends on the type of information and the purpose for which it is needed.

The newer privacy-moderation architecture uses the following retention controls:

• Masked privacy evidence is ordinarily retained for 30 days, except where an active appeal requires a hold.
• Privacy-deletion tombstones are retained for 90 days to help prevent previously deleted content from being captured again.
• Deleted message text and edit history are purged from the newer statistics archive.

Other records, including member history, general moderation records, community statistics, and token transactions, may be retained for as long as they remain necessary for their respective functions.

Botticelli may also retain the original questions submitted to its Knowledge Assistant in its own database. These records do not currently have a separate, uniform retention period.

Conversation data stored through OpenAI's Responses API is subject to the retention practices described in Section 6.

Retention practices for some legacy records are under review as the migration progresses.

Botticelli does not currently provide a single uniform retention period for every category of stored information.

9. Sharing Information

Botticelli may share information with service providers that support its operation, including hosting, database infrastructure, and AI processing services.

Information may also be made available to authorized server administrators or moderators when necessary for community management, safety, or review of moderation decisions.

Some limited information may be exchanged with other authorized community-management services where necessary to provide their functions.

Information may also be disclosed where required by applicable law.

10. Data Requests and Deletion

Members may request access to, correction of, or deletion of eligible personal information by contacting the moderation team of the Discord server where Botticelli operates or by using the privacy contact provided below.

Requests are handled manually. There is currently no self-service deletion interface.

Some information may need to be retained where necessary for an active moderation investigation, appeal, security matter, or other applicable legal requirement.

Requests are assessed according to the information held, its continuing necessity, and applicable obligations.

Where information has been stored by OpenAI through the Knowledge Assistant, Botticelli does not currently provide immediate deletion of those stored copies. Such information remains subject to OpenAI's applicable retention practices.

Requests concerning Botticelli's locally stored records are assessed separately.

11. Optional Participation and Moderation

Members may choose whether to participate in optional community activities such as challenges or rewards.

However, mandatory moderation and safety functions may process information from members regardless of whether they use Botticelli's commands.

Individual members cannot disable server-wide safety functions.

Server administrators may configure applicable collection exclusions and other settings where supported.

12. Security

Botticelli uses technical and administrative measures intended to protect information from unauthorized access, misuse, and unnecessary disclosure.

These measures include restricted access, moderation safeguards, and additional protections for sensitive information.

No system can guarantee absolute security.

13. Changes to This Policy

This policy may be updated when Botticelli's functionality, data practices, or applicable requirements change.

The latest version will be published with its revision date.

14. Contact

For privacy questions, data requests, or concerns about Botticelli, contact:

[email protected]

Requests will be directed to the responsible application administrator.